Association of medication adherence quality measures for diabetes, hypertension, and hyperlipidemia with cognitive decline.

Background: While diabetes, hypertension, and hyperlipidemia each are associated with increased risk of cognitive decline, little is known regarding how nonadherence to medications for these conditions is associated with cognitive decline risk. Methods: We identified patients enrolled in a Medicare Advantage Prescription Drug plan who were eligible for inclusion in the CMS Star Medication Adherence quality measures for diabetes, hypertension, and hyperlipidemia in 2018, 2019, and 2020. To achieve an adherence quality measure, patients had to meet 80% of the proportion of days for the medication. We used propensity score with inverse probability of treatment weighting to balance outcomes for baseline characteristics and logistic regression models to compare odds of cognitive decline outcomes across patient groups. Results: The study population of 99,774 individuals had a mean age of 71.0 years and was 49.1% female, 73.9% White, and 17.8% Black, with 62.0% living in an urban setting. Compared with patients who missed zero adherence measures, those who missed one measure had 23%-33% increased odds of cognitive decline (any decline OR = 1.23; dementia OR = 1.33; Alzheimer's disease OR = 1.27; all P values <0.01). Patients who missed 2-3 measures had 37%-96% increased odds of cognitive decline (any decline OR = 1.37; dementia OR = 1.58; Alzheimer's disease OR = 1.96; all P values <0.01). Patients who missed ≥4 adherence measures had the greatest odds of cognitive decline (any decline OR = 1.64; dementia OR = 2.05; Alzheimer's disease OR = 2.48; all P values <0.01). Conclusion: Not achieving CMS Star Medication Adherence quality measures for diabetes, hypertension, and hyperlipidemia therapies was associated with increased risk of cognitive decline outcomes.

Real-World Incidence and Management of Immune-Related Adverse Events from Immune Checkpoint Inhibitors: Retrospective Claims-Based Analysis.

PURPOSE: We assessed real-world spectrum and patterns of irAEs for patients treated with anti-PD(L)1 ICIs. METHODS: irAEs were defined using medical and pharmacy claims for patients enrolled in a Medicare Advantage Prescription Drug plan who initiated treatment with anti-PD(L)-1 and received ≥ 1 dose of therapy between 1 September 2014 and 28 February 2018. RESULTS: Treatment was discontinued for 46.6% of patients, and withheld and subsequently restarted for 10.3%. While toxicity profiles did not differ by age, RiskRx-V co-morbidity index was higher in patients with irAEs. CONCLUSION: These data underscore the needs for tailored irAE diagnostic and management pathways.

The impact of real-world cardiovascular-related pharmacogenetic testing in an insured population.

BACKGROUND: Pharmacogenomics is intended to help clinicians provide the right drug to the right patient at an appropriate dose. However, limited evidence of clinical utility has slowed uptake of pharmacogenomic testing (PGT). OBJECTIVE: To evaluate the impact of real-world cardiovascular (CV)-related PGT on clinical outcomes, healthcare resource utilisation (HCRU) and cost in a large, heterogeneous population. METHODS: Individuals with Medicare Advantage Prescription Drug, Medicaid, or commercial coverage between 1/1/2011 and 9/30/2015 and ≥1 atherosclerotic CV-related diagnosis were identified. Those with ≥1 claim for CV-related PGT were included in the test group (index date = 1st PGT claim) and matched 1:2 to controls without PGT. Individuals aged <22 or ≥90 years old on the index date, with <12 months continuous enrollment before and after the index date, or without an ASCVD-related diagnosis in the 12-month pre-index period were excluded. The primary outcome was occurrence of a major CV event during the 12-month post-index period. RESULTS: After adjustment, the PGT group was significantly more likely to experience ischaemic stroke, pulmonary embolism, deep vein thrombosis or a composite event compared with controls. Adjusting for baseline characteristics, HCRU was significantly higher for the test group across all measured outcomes except all-cause and ASCVD-related inpatient admissions. Median all-cause and ASCVD-related healthcare costs were significantly higher for the test group. CONCLUSIONS: Real world PGT in a large population did not improve outcomes. Tailoring medication therapy to each patient holds great promise for providing quality care but a deeper understanding of how widespread utilisation of PGT might impact objective health outcomes is needed.

Long-term stability of CMV DNA in human breast milk.

BACKGROUND: Human cytomegalovirus (CMV) is the leading cause of intrauterine and perinatal viral infection. The most common route of CMV transmission in newborns is through breastmilk and this can lead to infant morbidity and mortality. Breast milk that has been frozen for an extended period may need to be tested for CMV DNA to determine the source of infection. It has been a challenge for clinical laboratories to ensure the stability of CMV DNA in frozen breast milk for accurate viral load measurement. OBJECTIVES: To evaluate the stability of CMV DNA in breast milk by testing quantitative viral loads over a 28-day period for breast milk stored at 4 °C and a 90-day period for breast milk stored at -20 °C. STUDY DESIGN: Baseline viral loads were determined on day 0 and the samples stored at 4 °C underwent extraction and amplification at four time points, up to 28 days. The samples stored at -20 °C underwent extraction and amplification at five time points up to 90 days. Log10 values were calculated and t-test, Pearson's coefficient, and concordance correlation coefficient were calculated. RESULTS: There was no statistically significant difference between the time points by t-test, and correlation coefficients showed greater than 90% concordance for days 0 and 28 as well as days 0 and 90 at both storage temperatures tested. CONCLUSIONS: The concentration of CMV DNA in breast milk was stable for 28 days at 4 °C and 90 days at -20 °C as the concentrations did not differ significantly from the baseline viral loads.

Comparative Effectiveness of Rapid-Acting Insulins in Adults with Diabetes.

BACKGROUND: Although there are a variety of insulin products and new delivery modalities available, the absence of direct clinical and economic comparisons can make treatment planning and formulary decision making difficult. Direct comparisons between insulin aspart and insulin lispro from a large heterogeneous population are not available. OBJECTIVE: To assess differences in clinical outcomes, medication adherence, utilization, and total health care costs between aspart and lispro and vial versus pen modalities for administering these short-acting insulin analogs. METHODS: This retrospective cohort study used administrative claims data from the Humana Research Database to identify people with type 1 or type 2 diabetes and Medicare or commercial insurance (with medical and pharmacy benefits) who newly initiated rapid-acting insulin between January 1, 2008, and December 31, 2013, and were continuously enrolled during the 12-month baseline and 12-month follow-up periods. Generalized linear models were used to assess differences in costs and utilization. Logistic regression models measured the likelihood of having a hypoglycemic event, worsening diabetes complications, or a change in glycated hemoglobin (A1c). RESULTS: 8,189 patients included in the study were grouped by rapid-acting insulin product (aspart, n = 5,364, and lispro, n = 2,566) and modality (vial, n = 6,135, and pen, n = 2,054). There were no significant differences in the percentage of patients with a hypoglycemic event, new or worsening diabetes complications, or change in A1c, and there were no significant differences in adjusted total health care, medical and pharmacy costs, or emergency department visits between any of the product or modality comparisons. There was a significant difference in mean annual inpatient stays between lispro and aspart (adjusted mean = 2.24, 95% CI = 0.73-6.69, and adjusted mean = 2.65, 95% CI = 0.86-7.86, respectively; P < 0.001) and pen and vial cohorts (adjusted mean = 1.74, 95% CI = 0.56-4.99, and adjusted mean = 3.05, 95% CI = 1.01-9.08, respectively; P < 0.001). Adherence was similar for the lispro and aspart cohorts. Adherence was higher in the pen cohort (as measured by medication possession ratio ≥80%) compared with the vial cohort (adjusted odds ratio = 1.29, 95% CI = 1.12-1.50). CONCLUSIONS: This study provides a comprehensive assessment of outcomes and costs between 2 commonly used rapid-acting insulin products. Overall, there was little differentiation between products, although adherence improved significantly with pen devices. These findings may simplify decisions related to formulary options and choice of therapy. DISCLOSURES: No outside funding supported this study. Racsa and Ellis are employees of Comprehensive Health Insights, a subsidiary of Humana, and Saverno was employed with Comprehensive Health Insights at the time of this study. Meah is an employee of, and owns stock in, Humana. The authors have no financial disclosures or potential conflicts of interest to report. All authors contributed equally to study concept and design, data interpretation, and manuscript preparation. Racsa collected the data.

Comparing two tyrosine kinase inhibitors for treatment of advanced renal cell carcinoma in Medicare and commercially insured patients.

OBJECTIVE: The objective of this study was to compare treatment characteristics, survival and costs for sunitinib and pazopanib for advanced renal cell carcinoma (RCC) in a real-world setting. METHODS: Using claims data, this observational, retrospective cohort study selected individuals aged 19 to 89 years, with commercial or Medicare insurance, advanced RCC, and at least one pharmacy claim for sunitinib or pazopanib between 1 November 2009 and 31 December 2012. Treatment characteristics (treatment interruption, adherence, duration and discontinuation), survival, and costs were measured up to 12 months. Statistical models were adjusted for age, gender, geographic region, race, and RxRisk-V score. RESULTS: At baseline, pazopanib patients exhibited significantly worse health status indicators (RxRisk-V score, number of pharmacy claims, and pre-index total healthcare costs) than sunitinib patients. There were no differences in treatment characteristics or survival. Index medication costs (mean difference $5580, p = 0.03, adj p = 0.05) and total healthcare costs (mean difference $12,192, p = 0.09, adj p = 0.07) trended higher with sunitinib. Patients non-adherent with sunitinib incurred significantly higher total costs compared to patients non-adherent with pazopanib (mean difference $17,680, p = 0.04, adj p = 0.01). CONCLUSIONS: Mortality data and proxy variables for treatment effectiveness indicate comparable clinical value for both medications. Sunitinib treatment trended towards higher index medication and total healthcare costs despite higher pre-index total costs and worse health status indicators at baseline with pazopanib. Non-adherence with sunitinib was associated with significantly higher total healthcare costs, which may indicate differences in tolerability between the two agents and requires further investigation.