The Patient Protection and Affordable Care Act (ACA) allows individuals and small businesses to purchase private health insurance through competitive marketplaces. Off-exchange plans are ACA compliant, but sold outside the marketplace, and little is known about the characteristics of enrollees of these plans. This analysis examines member severity and annual cost of enrollees of individual versus small group plans. This study used a national sample of claims data for 2.0m individual and 1.7m small group members enrolled in an off-exchange plan in 2017. Cross sectional descriptive analyses were conducted to examine differences between enrollees of individual and small group plans. Individual plan members were much more likely to enroll in a Silver plan (62% versus 36% small group) while small group members were most likely to enroll in a Gold plan (41% versus 7% individual). Average risk scores were higher among individual members by 20% Silver, 27% Gold, and 66% Platinum. Average PMPM costs were also higher in individual members by >2x Platinum, 36% Gold, and 8% Silver. Cost-risk ratios were also higher in individual members by 23% Platinum and 7% Gold, but were 10% lower in the largest group of Silver members. Risk scores were similar among individual and small group Bronze members, but costs were 33% lower in individual enrollees, and mean cost-risk ratios were 23% lower. Overall, the cost-risk ratio was 7% lower in individual plan members compared to those in small group plans. In general, off-exchange members enrolled in individual plans were less costly (relative to risk score) than those enrolled in small group plans. These findings provide insight into the enrollment patterns and relative risk and cost of individual and small group off-exchange enrollees. The results provide new insights into which individual plan members are more likely to be profitable for off-exchange health plans.

To compare selected healthcare utilization and medication adherence outcomes for Medicare Advantage (MA) beneficiaries with diabetes enrolled in diabetes-related Chronic Condition Special Needs Plans (C-SNPs) versus similar enrollees in non-Special Needs Plan (SNP) MA plans. This study used claims from a large nationally representative claims database augmented by linking patients to publicly available information on SNP plan participation and specialty, if any, and social risk factor characteristics at the near neighborhood level (nine-digit ZIP) on median household size and mean household income. The sample consisted of MA beneficiaries in the study dataset enrolled either in a diabetes-related C-SNP (N=5,100) or no SNP (N=449,581) with at least one documented diabetes diagnosis or at least one fill for an antidiabetic drug in 2015. Five quality measures were computed to compare the diabetes C-SNP patients to non-SNP patients with diabetes: percent of enrollees with at least one primary care physician (PCP) visit, HbA1c test, inpatient admission or readmission; and percent of days covered with antidiabetic medication. A risk adjustment model of each outcome on the non-SNP population was calibrated and applied to the C-SNP population to identify any "C-SNP effect". MA beneficiaries enrolled in diabetes-related C-SNPs appear to have experienced improved outcomes compared to beneficiaries enrolled in non-SNP plans. PCP visits were 22 percent higher, diabetes testing was 10 percent higher, diabetes medication adherence was 6 percent higher, and inpatient admissions and readmissions were more than 30 percent lower for diabetes C-SNP enrollees compared to those enrolled in non-SNP plans. The evidence suggests that outcomes are better for Medicare beneficiaries who are enrolled in diabetes-related C-SNPs than would have been realized had those individuals enrolled in non-SNP MA plans. These findings are consistent across multiple domains of care: primary care, inpatient utilization, and diabetes management.

To compare multiple sclerosis (MS) related healthcare use within one year of initiating first and second generation disease modifying therapies (DMTs) among individuals diagnosed with MS in the U.S. A large national sample of patient-level administrative healthcare claims data were used for this analysis. MS patients aged 18 years and over with a new prescription fill for a first or second generation DMT between 2013 and 2015 were evaluated. First generation DMTs included interferon beta-1b and glatiramer acetate (GA). Second generation DMTs included teriflunomide, dimethyl fumarate, natalizumab, novantrone, interferon beta-1a, peginterferon beta-1b, and fingolimod. Patients were eligible if they were continuously enrolled in a health plan with pharmacy and medical coverage for at least 6 months before and 1 year after initiation of therapy. Three types of healthcare use were examined: MS-related hospitalizations and relapse events (inpatient and outpatient). Multivariate Poisson and Zero-inflated Poisson models with robust standard errors were used to estimate the association between MS-related healthcare use and type of DMT. All models adjusted for age, gender, Charlson index, geographic region and payer type. During the study period, 18,946 individuals with a MS diagnosis initiated a DMT. Of those, 77.3% were female, 66.3% were ages 40-64 years, and 53.3% had commercial health insurance. Almost two-thirds (62.9%) were on second generation DMTs. Second generation DMT users were found to have fewer MS-related hospitalizations compared to first generation DMT users within one year of initiating therapy (adjusted risk ratio (ARR)= 0.91 (95% confidence interval (CI): 0.88-0.97)). Differences in inpatient and outpatient relapses were not observed by DMT type. This study suggests that second generation DMTs are associated with reduced MS-related hospitalizations within one year of initiating therapy. Studies examining a longer treatment timeframe and additional healthcare outcomes are warranted to confirm these findings.

To determine factors associated with having one or more prescription drug fills (PDFs) in members of qualified health plans (QHPs). This study used a large nationally representative administrative claims database supplemented with socioeconomic and community resource data. The sample consisted of 1,823,677 members of QHPs who were enrolled for ≥10 months in 2014. Logistic regression was used to determine factors associated with having PDFs. A total of 1,230,981 (67.5%) members had ≥1 PDF during the study period. Living in a neighborhood with higher education levels (i.e., percent population with at least bachelor’s degree) was associated with higher likelihood of having PDFs (OR: 1.04-1.12; reference: 0-12%). Lower median household income was associated with lower likelihood of having PDFs (OR: 0.73-0.90; reference $100,000+). Members from neighborhoods with a higher proportion of non-white race/ethnicities were less likely to have PDFs (non-Hispanic black, OR=0.80; Hispanic, OR=0.95; Asian, OR=0.58). Compared to Silver plan members, members of Gold plans were more likely to have PDFs (OR=1.07); but members of Platinum, Bronze, and Catastrophic plans were less likely to have PDFs (OR: 0.95, 0.65 and 0.83, respectively). Individual plan enrollment was associated with higher likelihood of having PDFs (OR=1.10), while on-exchange plan enrollment was associated with lower likelihood (OR=0.87). Other factors associated with lower likelihood of having PDFs were being male (OR=0.62); younger age (OR: 0.55-0.71; reference: 31-65); and living in a non-metropolitan area (OR: 0.89-0.95). This study provides evidence that the likelihood of having PDFs varies across member and plan characteristics, even after controlling for patient severity. Assuring appropriate treatment and adherence to medications for chronic conditions is inherent to achieving good health outcomes, thus it is important that QHPs identify and understand factors associated with lower likelihood of filling prescriptions in order to develop targeted interventions to achieve the most optimal outcomes in their members.

To evaluate the risk of adverse events associated with the use of oral antidiabetic agents in middle-aged patients with type II diabetes mellitus (T2DM). This retrospective-cohort study used propensity score matching to identify patients in a large nationally representative administrative claims database. The sample consisted of T2DM patients aged 30-64 years who were new users of Sodium-Glucose Cotransporter 2 (SGLT2) inhibitors or Dipeptidyl Peptidase-4 (DPP-4) inhibitors between 2013 and 2015. Patients were excluded if they: (a) were prescribed polytherapy of diabetic drugs within seven days of the index prescription; (b) were not continuously enrolled during the 12 months before initiation of treatment; (c) did not have medical or pharmacy benefits; or (d) had a diagnosis of a serious medical condition (i.e., active cancer, HIV/AIDS, organ transplant, and ESRD) at baseline. Cox proportional hazards models were performed and eligible patients were followed to the time of adverse event or up to 180 days after the index date. They were censored if switched to another drug agent, disenrolled from the health plan, or end of study period. A total of 11,734 T2DM patients were identified as new users of SGLT2 or DPP-4 inhibitors after matching (female = 53.7%, age = 54.0[± 7.6], history: mild liver disease = 8.0%, retinopathy = 11.4%, neuropathy = 16.2%). Compared to patients treated with DPP-4 inhibitors, those with SGLT2 inhibitors had a significantly lower risk of cardiovascular disease (HR = 0.766, 95% CI: 0.633-0.927) after controlling for potential confounders. There were no significant differences in the risk of developing ketoacidosis (DKA) or osteoporotic fractures between treatments. This study provides new evidence that use of SGLT2 inhibitors decreases the risk of cardiovascular disease in middle-aged patients with T2DM. This suggests that DPP-4 inhibitors should be prescribed with caution in this patient population, particularly in patients with other risk factors for cardiovascular events.

To examine the relationship between demographic and socio-economic factors and medication adherence (MA). This study used a large nationally representative administrative claims database supplemented with socio-economic and community resource data. The sample consisted of Medicare Advantage members enrolled in 2013. Multivariate logistic regression was used to determine factors associated with 3 MA measures from the CMS Five-Star Quality Rating System: Cholesterol (Statins) (MA-C), Diabetes Medications (MA-D) and Hypertension (RAS Antagonists) (MA-H). A total of 764,581 members from 44 health plans qualified for at least 1 of the measures (MA-C=68.0%, MA-D=72.2% and MA-H=74.0%). For all 3 measures, MA was significantly lower for members who were: younger and had a disability (OR: 0.54-0.95); African-American (OR: 0.64-0.66) (and, for MA-C and MD-D only, Hispanic (OR: 0.79-0.90)). MA was lower for those who resided in an area with higher percent of population below federal poverty level (OR: 0.77-0.91), and was significantly higher for members who resided in an area with higher home ownership (OR: 1.05-1.08) or higher education level (OR: 1.04-1.07). After controlling for socio-economic and clinical (measured by number of different medications the member was taking) characteristics, MA was significantly higher for dual eligible members, with full benefit duals being more adherent (OR: 1.09-1.16) than partial duals (OR: 1.07-1.09). In addition, sub-group analyses suggested that non-duals who were low income actually had lower adherence than duals who were poor (but have more benefits due to dual status). Primary care shortage area was not a significant risk factor for MA. MA is significantly associated with demographic and socio-economic factors. Health plans serving a high proportion of disadvantaged members may be providing better quality of care than their MA measure results suggest. Specifically, dual members had higher adherence rates than members with similar characteristics who did not receive Medicaid benefits.

To identify factors associated with emergency room (ER) and primary care provider (PCP) visits by members of qualified health plans (QHPs). This study used a large nationally representative administrative claims database supplemented with neighborhood level socioeconomic data. The sample consisted of 1,823,677 QHP members enrolled for ≥10 months in 2014. Logistic regression was used to determine factors associated with ER and PCP visits. A total of 15.8% of the study population members had ≥1 ER visit and 72.3% had ≥1 PCP visit. Factors associated with lower likelihood of ER visit were higher percentage of neighborhood population with bachelor’s degree or higher (ORs: 0.90-0.98; reference 0-12%); non-metropolitan area (OR: 0.72-0.90); enrolled in Bronze (OR=0.82) or Catastrophic (OR=0.9) metal level plan; and enrolled on-exchange (OR=0.96) compared to off-exchange. Factors associated with higher likelihood of ER visit were lower median household income (ORs: 1.03-1.16) or higher proportion of non-Hispanic blacks (OR=1.19) neighborhood; and enrolled in Gold (OR=1.05) or individual group plan (OR=1.08). Factors associated with lower likelihood of PCP visit were lower median household income (ORs: 0.73-0.89; reference $100,000+) or higher proportion of Hispanics (OR=0.80) or non-Hispanic blacks (OR=0.86) neighborhood; non-metropolitan area (ORs: 0.92-0.96), enrolled in Bronze (OR=0.70) or Catastrophic (OR=0.90) plan; and enrolled on-exchange (OR=0.96). Factors associated with higher likelihood of PCP visit were higher percentage of neighborhood population with bachelor’s degree or higher (ORs: 1.03-1.07); enrolled in Gold (OR=1.09) or individual group plan (OR=1.22). The effects of socioeconomic characteristics on the likelihood of having an ER visit were the opposite of their effects on likelihood of having a PCP visit. However, the effects of plan characteristics on the likelihood of ER or PCP visits were in the same direction.

To evaluate the impact of positive airway pressure (PAP) treatment on healthcare utilization among patients with obstructive sleep apnea (OSA). This retrospective-cohort study was conducted using a large nationally representative administrative claims database. Patients were included in the study if they: (1) were ≥65 years; (2) had a polysomnography (PSG) diagnostic test between 2008 and 2013; (3) had ≥2 medical claims with OSA diagnosis (ICD-9-CM code 327.23) within 1 year after the first PSG test; (4) were newly treated (index) or never treated with PAP after OSA diagnosis; and (5) were continuously enrolled in a plan with medical benefits for 12 months prior and 24 months after the index date. Patients were excluded if they: (1) had claims indicating the use of PAP device, PAP-related supplies, or PAP management prior to the first PSG diagnostic test date; or (2) had a diagnosis of other primary respiratory conditions at baseline. Multivariate logistic regression and negative binomial models were estimated to examine the impact of PAP treatment on all-cause hospitalizations, sleep apnea-related hospitalizations and emergency room (ER) visits. A total of 24,420 patients (mean age: 72.0±5.1; 58% males) were identified, of which 89% used PAP. Mean Charlson score was not significantly different between patients with PAP and those without PAP (1.44±1.33 vs. 1.46±1.36, p=0.88). After controlling for potential confounders, patients with PAP were less likely to have an all-cause hospitalization than those without PAP (OR=0.90, 95% CI: 0.82-0.98, p=0.02). Considering frequency, those with PAP had significantly fewer all-cause hospitalizations (RR=0.86, 95 CI: 0.80-0.93, p<0.01), sleep apnea-related hospitalizations (RR=0.86, 95 CI: 0.77-0.97, p=0.01), and ER visits (RR=0.90, 95 CI: 0.85-0.96, p<0.01). This study provides new evidence that the use of PAP devices decreases the risk of hospitalizations and ER visits in elderly patients with OSA.

To examine differences in the likelihood of receiving Breast Cancer Screening (BCS) among Medicare Advantage (MA) dual eligibles (DE) enrolled in Special Need Plans (D-SNPs), DE in non-SNPs (non-SNP DEs), and non-DE members. This study used a large nationally representative administrative claims database supplemented with socioeconomic and community resource data. The sample consisted of female MA enrollees aged 50-74 years in 2013. The outcome of interest was an indicator of receiving BCS based on the measure definition in the Healthcare Effectiveness Data and Information Set. Generalized linear mixed model was used to compare the likelihood of receiving BCS in the three groups after controlling for confounding factors (i.e., demographics, comorbidities, socioeconomic characteristics and community healthcare resources) and accounting for unmeasured plan characteristics as a random component. A total of 258,807 MA members were included in the study (non-DE: 76.0%, non-SNP DEs: 10.1%, D-SNP: 13.9%). BCS rates were significantly different across all three groups (p-value<0.0001). The non-DE population had higher rates (77.3%) compared to both non-SNP DE (72.9%) and D-SNP (76.3%). The model revealed there was no significant difference in the likelihood of receiving BCS between D-SNPs and non-DEs (OR: 1.1, p-value= 0.33); however the likelihood of receiving BCS was lower in non-SNP DEs than in both non-DEs (OR: 0.81, p-value<0.0001) and D-SNPs (OR: 0.76, p-value= 0.0002). The probability of receiving BCS was lower in dual members not in a D-SNP plan than in duals enrolled in D-SNP plans and non-DE populations. There was no significant difference in the probability between D-SNP and non-DE populations. The findings indicate that SNP plans produce better results in dual members compared to duals not in a SNP plan. This provides evidence of the value of SNP plans in achieving better outcomes for the vulnerable DE MA population.

To identify factors associated with disease modifying antirheumatic drugs (DMARDs) receipt among Medicare beneficiaries enrolled in Medicare Advantage (MA) plans. This study used a large nationally representative administrative claims data, supplemented by new sources of socioeconomic and community resource data (i.e., market source (at zip+4 levels) and area health resource files) in addition to CMS published contract information and Monthly Membership Report. The sample consisted of MA members (≥18 years) who were diagnosed with rheumatoid arthritis (RA) during 2013. The generalized linear mixed model was used to determine factors associated with DMARD receipt. A total of 12,835 RA patients were identified, of which 9,850 (76.74%) received DMARD. Factors associated with less likelihood of receiving DMARD were male gender (OR: 0.72), increasing age (80-84 years, OR: 0.67; ≥85 years, OR: 0.45; vs. 18-54 years), more comorbidities (Charlson Comorbidity Index, OR: 0.96; HCC Risk Score, OR: 0.89), living in South Atlantic region compared to Mid-Atlantic (OR: 0.76), and percent households with 1st Individual who completed college (1-15% vs. 0%, OR: 0.82). Factors associated with more likelihood of DMARD receipt were Hispanic compared to White (OR: 1.38), use of glucocorticoids (OR: 2.22), living in New England region compared to Mid-Atlantic (OR: 1.82), median household income ($50,000-$74,999, OR: 1.31; $75,000 - $99,999, OR: 1.34; $100,000+, OR: 1.42; vs. $0-$15,000), end-stage renal disease (OR: 2.77), enrolled in employer group waiver plan (OR: 1.40), MA plans market penetration rate in service area (3-4%, OR: 1.26; 5-8%, OR: 1.45; 9-100%, OR: 1.60; vs. 0-2%). The probability of receiving DMARD is correlated with sociodemographic, clinical, health plan and community characteristics. This study provides new evidence that can be used to identify subgroups of members to effectively target interventions to improve arthritis management in the Medicare managed care population.

The CMS Five Star Rating system for Medicare Advantage (MA) informs beneficiaries about plan performance and determines Quality Bonus Payments. High Risk Medication (HRM) use is a triple weighted measure defined as the percent of beneficiaries aged 65+ who received two or more fills for a drug with high risk of serious side effects in the elderly. This analysis evaluates differences in HRM use between dual eligible (DE) and non-DE members and examines the contribution of socio-demographic and clinical characteristics to observed disparities. The study used a nationally representative administrative claims database of 1.5 million MA members in 2011-2012 (measured in member years): 232,273 DE (female: 66.3%; average age 75.4) and 1,251,145 non-DE (female 57.4%; average age 75.2). The Linear Probability Model (LPM) and Blinder-Oaxaca decomposition techniques with Neumark weighting formula were performed. HRM rate was 32.2% higher in DEs (16.8% vs. 12.7%). The decomposition analysis found member characteristics accounted for only 48% of the performance gap (“explained gap”); 52% was attributed to differential effects of member characteristics on HRM use (“unexplained gap”). The Charlson Severity Score indicates more complex comorbidities in DEs (2.34 vs. 1.77) and explained 70.6% of the difference in HRM use; disability as original reason for entitlement explained 23.3%. Members of a Preferred Provider Organization (PPO) and older members were less likely to use HRMs, while females and low income members were associated with higher use. This study provides information about the contribution of socio-demographic and clinical characteristics to higher use of HRMs in DEs, and can support targeted interventions to reduce the performance gap. The analysis further demonstrates that more than half the disparity is not explained by the member characteristics evaluated and points to the need for further research to understand the factors behind the unexplained DE gap.

To evaluate the relationship between asthma-related health care utilization with two different types of asthma treatments (inhaled Fluticasone Propionate (FP) and fixed dose combination of Fluticasone Propionate and Salmeterol (FPS)) among a commercially insured adolescent and adult population. This retrospective cohort study used propensity-score matching to identify patients in a large nationally representative administrative claims database. The sample consisted of patients aged 12 to 64 who had a new prescription fill (index) for FP or FPS after asthma diagnosis (ICD-9-CM 493.XX) between 2006 and 2012. Patients were continuously enrolled in a plan with pharmacy and medical coverage for at least 12 months before and after treatment initiation. Patients were excluded if they: (a) used FP or FPS within 90 days prior to index; (b) had a diagnosis of chronic obstructive pulmonary disease (ICD-9-CM 490.XX-492.XX, 494.XX, 496.XX); or (c) used other asthma controller medications after index. Eligible patients were followed up to 12 months after initiation. A multivariate logistic regression model was used to observe the association between asthma-related health care utilization in the two cohorts. The study population included 8,059 patients treated with FP (female=57.0%, average age=33.9 (± 17.6), hospitalization rate=1.4%, ED rate=5.0%) and 8,059 patients treated with FPS (female=57.5%, average age=33.9 (± 17.4), hospitalization rate=1.8%, ED rate=5.5%). The multivariate logistic regression model revealed FPS users had no significant difference in likelihood of asthma-related hospitalizations (OR=1.26; 95% CI 0.98-1.62, p-value=0.0667) or ED visits (OR=1.13; 95% CI 0.98-1.3, p-value=0.0884) compared to FP users. This large retrospective analysis found that users of FP versus FPS have statistically similar rates of asthma-related health care utilization at the 95% level. However, the large positive odds ratios, significant at the 90% level, indicate a need for additional research to investigate comparative effectiveness of the two therapies.

To evaluate the relative likelihood of severe cutaneous reactions (Steven-Johnson Syndrome, toxic epidermal necrolysis), aseptic meningitis, and organ dysfunction (pancreatitis, hepatotoxicity) associated with antiepileptic drugs (AEDS) in children (aged 2-18 years) as compared to adults (aged 19+) with epilepsy. This retrospective cohort study analyzed patients in a large nationally representative administrative claims database between 2006 and 2011. The sample consisted of Medicaid, Medicare, and Commercial patients with a diagnosis of epilepsy (ICD-9-CM 345.X) who were continuously enrolled with pharmacy benefits for 6-months and had no prior AEDS fills (new users). Multivariate analysis (logistic regression) was used to follow eligible patients from index date to first adverse event or up to six months after AED initiation. The study population included 1,803,871 new users of AEDS. Children comprised 7.9% (N=142,874, female=50.7%, age=12.4 [± 4.7]) and adults 92.1% (N=1,660,997, female=64.6%, age=52.8 [± 17.7]). Compared to adults, children were significantly less likely to experience organ dysfunction (OR=0.228, 95% CI: 0.20-0.26, p<0.0001), but more likely to have severe cutaneous reactions (OR=1.19, 95% CI: 1.14-1.25, p<0.0001) after controlling for potential confounders (gender, region, type of medication). There was no significant age effect on risk of aseptic meningitis. Female adults had a statistically lower likelihood of organ dysfunction compared to males (OR=0.75, 95% CI: 0.72-0.77, p<0.0001), but there was no gender effect for risk of organ dysfunction in children. Females in both groups had a higher likelihood of developing cutaneous reactions compared to males (adults: OR=1.28, 95% CI: 1.25-1.31, p<0.0001; children: OR=1.20, 95% CI: 1.15-1.26, p<0.0001). Adverse effects resulting from treatment with AEDS are an important consideration in evaluating epilepsy treatment options. This study provides new information about the comparative risks of AEDS that can be used to guide optimal prescribing practices for patients with epilepsy.

To evaluate the relative likelihood of cardiac events, pancreatitis, pneumonia, and death associated with use of valproate (V) as an adjunct “off label” therapy to antipsychotics (AP) in the elderly population with schizophrenia. This retrospective cohort study utilized a propensity-score matching technique and a new user design to identify patients in a large nationally representative administrative database. The sample included Medicaid, Medicare, and Commercial patients aged 65+ with a diagnosis of schizophrenia who were continuously enrolled with pharmacy coverage for 12-months before treatment initiation. Patients with a diagnosis of bipolar, epilepsy or migraine were excluded. The AP cohort was identified by the first prescription fill with no prior use in previous 12-months. The combination therapy cohort (V+AP) was identified by the first prescription fill for V and an AP fill within 30 days. Multivariate survival analysis was used to follow patients to first adverse event or up to six years after index date to assess relative risk of adverse events. The study population included 1,348 patients treated with V+AP (female=52.6%, age=79.3 (± 6.9); history: liver disorder=5.9%, congestive heart failure=28.5%, peripheral arterial disease=32.2%) and 1,348 AP only patients (female=50.3%, age=79.1 (± 6.8); history: liver disorder=14.3%, congestive heart failure=27.9%, peripheral arterial disease=34.3%). Patients treated with AP only had a significantly lower risk of cerebrovascular disease (HR=0.65, 95% CI: 0.51-0.85, p=0.01) after controlling for potential confounders. Adding V did not increase risk of other outcomes evaluated. Clinical trials have failed to provide evidence of the efficacy of V as an adjunct therapy to AP to treat schizophrenia, but “off-label” prescribing of V+AP remains high. This study provides new evidence to inform prescribing practices in elderly patients with schizophrenia. The increased risk of stroke must be weighed against any incremental benefit to the patient that the addition of V may provide.