Best practices in the real-world data life cycle.

With increasing digitization of healthcare, real-world data (RWD) are available in greater quantity and scope than ever before. Since the 2016 United States 21st Century Cures Act, innovations in the RWD life cycle have taken tremendous strides forward, largely driven by demand for regulatory-grade real-world evidence from the biopharmaceutical sector. However, use cases for RWD continue to grow in number, moving beyond drug development, to population health and direct clinical applications pertinent to payors, providers, and health systems. Effective RWD utilization requires disparate data sources to be turned into high-quality datasets. To harness the potential of RWD for emerging use cases, providers and organizations must accelerate life cycle improvements that support this process. We build on examples obtained from the academic literature and author experience of data curation practices across a diverse range of sectors to describe a standardized RWD life cycle containing key steps in production of useful data for analysis and insights. We delineate best practices that will add value to current data pipelines. Seven themes are highlighted that ensure sustainability and scalability for RWD life cycles: data standards adherence, tailored quality assurance, data entry incentivization, deploying natural language processing, data platform solutions, RWD governance, and ensuring equity and representation in data.

Fournier's gangrene in patients with type 2 diabetes using second-line antidiabetic medications.

Cases of a rare but serious infection called Fournier's gangrene have been reported with sodium-glucose co-transporter-2 inhibitors (SGLT-2i). To evaluate the safety signal in a population of patients with type 2 diabetes, we used administrative claims data from Horizon Blue Cross Blue Shield of New Jersey from 2014 through 2017 to estimate incidence rates of Fournier's gangrene or necrotizing fasciitis of the perineum among patients treated with a second-line antidiabetic drug. We found very low incidence rates of Fournier's gangrene or necrotizing fasciitis. While we found no indication of an increased risk among SGLT-2i users compared with similar patients treated with other second-line antidiabetic medications, the small number of events yielded wide confidence intervals.

Actionable Real-World Evidence to Improve Health Outcomes and Reduce Medical Spending Among Risk-Stratified Patients with Diabetes.

BACKGROUND: Type 2 diabetes mellitus (T2DM) is a chronic condition with a high economic burden as well as drug treatments that have not all demonstrated effects on longevity. Managed care organizations want to improve health outcomes in these complex patients but lack actionable evidence to make informed decisions on which therapies are most effective among their members and may also control total health care spending. OBJECTIVE: To produce actionable evidence by identifying antidiabetic treatments that are effective and may reduce total cost of care in various risk groups of patients with T2DM, using insurance claims data that includes medical claims and pharmacy dispensing data among members of Horizon Blue Cross Blue Shield of New Jersey with T2DM. METHODS: We identified patients with T2DM in longitudinal claims data from Horizon between 2014 and 2017 with demographic and enrollment information, inpatient and outpatient diagnoses and procedures, and pharmacy dispensing. Outcomes included myocardial infarction, heart failure (HF), stroke, percutaneous revascularization, health care services utilization, and plan costs (i.e., medical, pharmacy, and total cost of care). After propensity score decile adjustment on over 20 covariates, we evaluated the effectiveness and safety of second-line antidiabetic treatment that included sodium-glucose co-transporter-2 (SGLT-2) inhibitors, sulfonylureas (SUs), dipeptidyl peptidase-4 (DPP-4) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists. RESULTS: Among 115,308 members with T2DM, the most common comorbidities were cardiovascular risk factors, including hyperlipidemia (56%), hypertension (50%), and existing cardiovascular disease (CVD; 55%). Among members receiving dual antidiabetic treatment (n = 20,204), the most prevalent treatments were metformin plus the following second-line medications: SUs (42%), DPP-4 inhibitors (29%), SGLT-2 inhibitors (10%), or GLP-1 receptor agonists (3%). Approximately 20% of members accounted for 79% of total cost of care, with an average of $9,605 per member per year (PMPY). Compared with SU initiation and after propensity score decile adjustment, new users of SGLT-2 inhibitors had a reduced risk for HF hospitalization (HR = 0.35, 95% CI = 0.13-0.89), hypoglycemia, albuminuria, microvascular disease, and metabolic failure. Among SGLT-2 inhibitor initiators with established CVD, the savings in total cost of care compared with SU initiators was $5,520 per member over an average treatment duration of 6 months and an approximate savings of $11,000 PMPY if patients persisted on treatment for 12 months. CONCLUSIONS: In the Horizon membership, we confirmed that SGLT-2 inhibitors reduce HF hospitalizations, resulting in reduced medical spending and savings in total cost of care. Regulatory-grade analytics of local data provided the confidence to encourage increased SGLT-2 inhibitor use to produce better outcomes and save total cost of care despite higher pharmacy spending. DISCLOSURES: This research did not receive outside funding; however, Aetion has since begun a contractual relationship with Horizon Blue Cross Blue Shield of New Jersey. Garry, Petruski-Ivleva, Cheever, and Rajan are employees of and have stock options in Aetion, a company that makes software for the analysis of real-world data. Eapen was an employee of Aetion during the implementation of this study. Rassen is an employee of and has ownership interest in Aetion. Murk is a consultant to Aetion of which he owns equity. Schneeweiss is a consultant to WHISCON and to Aetion, of which he also owns equity. He is the principal investigator of investigator-initiated grants to the Brigham and Women's Hospital from Bayer, Genentech, Boehringer Ingelheim, and Vertex. Gambino is an employee and officer at Horizon Blue Cross and Blue Shield of New Jersey. He was recently appointed to a board observer position at Aetion, as Horizon has small equity interest in Aetion. Jan is an employee of Rutgers State University and Horizon Blue Cross Blue Shield of New Jersey and has no conflict of interest or association with Aetion or any pharmaceutical company. Jang and Rubin are employees of Horizon Blue Cross and Blue Shield of New Jersey and have no conflict of interest or association with Aetion. This work was presented as a poster at AMCP Nexus 2018, October 22-25, 2018, in Orlando, FL; as part of a continuing education session at the AMCP Managed Care & Specialty Pharmacy 2019 Annual Meeting in San Diego, CA, March 25-28, 2019; as invited podium presenter at the Blue Cross Blue Shield 2019 National Summit conference in Grapevine, TX, April 29-May 2, 2019; and was accepted for a podium presentation at the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 2019 annual conference in New Orleans, LA, May 18-22, 2019, where it won an award for Best Podium Presentation.

Implications of early treatment among Medicaid patients with Alzheimer's disease.

OBJECTIVE: The objective of this study was to examine the effect of treatment timing on risk of institutionalization of Medicaid patients with Alzheimer's disease (AD) and to estimate the economic implications of earlier diagnosis and treatment initiation. METHODS: New Jersey Medicaid claims data (1997-2009) were used retrospectively to study the effect of treatment on time to institutionalization. Observed Medicaid payments were used to calculate savings from delayed institutionalization, adjusting for cost offsets resulting from concurrent changes in use of other medical services. RESULTS: Initiation of existing therapies at earliest symptomatic onset is predicted to delay institutionalization by 91 days, reducing Medicaid costs by $19,108/institutionalized patient. Incorporating an 18.5% cost offset from increased use of other medical services as well as drug costs associated with earlier treatment results in net savings of $12,687/patient. Projected annual Medicaid savings exceed $1 billion. CONCLUSION: Earlier treatment leads to a small delay in institutionalization among AD patients, resulting in significant costs savings to Medicaid.

Pre-diagnosis excess acute care costs in Alzheimer's patients among a US Medicaid population.

BACKGROUND AND OBJECTIVES: Prior research has documented that Alzheimer's disease (AD) is associated with increased costs from comorbid conditions. However, little is known about medical resource utilization and costs among AD patients prior to the onset of cognitive symptoms. This study estimates excess acute care costs among Medicaid AD patients in the year prior to diagnosis. STUDY DESIGN: Administrative claims data for New Jersey Medicaid patients over the period 1997-2010 were retrospectively analyzed. The study focused on non-institutionalized AD patients and examined their medical costs compared with matched controls with no dementia over the 12 months prior to their preliminary diagnosis. Costs reflect amounts reimbursed by Medicaid to medical service providers, reported in 2010 US dollars. RESULTS: The study sample included 11,536 AD patients who were matched to controls. Average age was 76 years, and 76.2 % were female. Compared with matched controls, total medical costs over the 12-month pre-index period were US$ 5,549 higher among AD patients (US$ 14,977 vs. US$ 9,428, p < 0.001), of which US$ 3,321 (p < 0.001) was due to outpatient services. Home care and medical daycare services accounted for US $1,442 (p < 0.001) of the difference. Emergency department visits and inpatient care accounted for only a small fraction of the excess costs. CONCLUSIONS: Compared with controls, Medicaid AD patients incurred higher acute care costs in the 12 months prior to their preliminary diagnosis, suggesting room for beneficial interventions and better disease management should earlier diagnosis become possible. These findings may be especially relevant in light of new criteria facilitating earlier diagnosis of AD.

The economic burden of metastatic breast cancer: a U.S. managed care perspective.

This study was performed to quantify the economic burden and identify drivers of direct costs of mBC. In a retrospective study of a de-identified administrative claims database of privately insured patients, women between 18 and 64 years of age were included if they had at least one claim with a diagnosis of breast cancer and subsequently one or more claims with a diagnosis of secondary malignancy between January 1, 2003 and December 31, 2009. The study sample was further classified into the following subgroups: (1) Endocrine therapy, (2) HER-2 targeted therapy, (3) Concomitant HER-2 targeted and endocrine therapy, (4) Cytotoxic chemotherapy, and (5) No-systemic therapy. Costs for medical resource utilization were calculated on a per patient per month (PPPM) basis. A total of 7,698 mBC patients were identified from 2003 to 2009 with an average age at index of ~52 years, and average follow up of 2.2 years. The average total direct medical costs for 7,698 mBC patients were $9,788 PPPM. Outpatient costs accounted for the majority of overall PPPM costs. Examining the five different mBC therapeutic subgroups revealed that patients who received no-systemic therapy had the highest costs at $13,926 PPPM, while patients who received systemic endocrine therapy had the lowest costs at $5,303 PPPM. This study demonstrated that mBC is associated with substantial healthcare costs in a non-Medicare patient population. Assuming average PPPM costs of $9,788 and an average life expectancy of 2.2 years, the total average expenditure to treat mBC would be ~$250,000 per patient.

Economic impact of the clinical benefits of bariatric surgery in diabetes patients with BMI ≥35 kg/m².

The medical costs for a type 2 diabetes patient are two to four times greater than the costs for a patient without diabetes. Bariatric surgery is the most effective weight-loss therapy and has marked therapeutic effects on diabetes. We estimate the economic effect of the clinical benefits of bariatric surgery for diabetes patients with BMI ≥ 35 kg/m². Using an administrative claims database of privately insured patients covering 8.5 million lives 1999-2007, we identify obese patients with diabetes, aged 18-65 years, who were treated with bariatric surgery identified using Healthcare Common Procedure Coding System codes. These patients were matched with nonsurgery control patients on demographic factors, comorbidities, and health-care costs. The overall return on investment (RoI) associated with bariatric surgery was calculated using multivariate analysis. Surgery and control patients were compared postindex with respect to diagnostic claims for diabetes, diabetes medication claims, and adjusted diabetes medication and supply costs. Surgery costs were fully recovered after 26 months for laparoscopic surgery. At month 6, 28% of surgery patients had a diabetes diagnosis, compared to 74% of control patients (P < 0.001). Among preindex insulin users, insulin use dropped to 43% by month 3 for surgery patients, vs. 84% for controls (P < 0.001). By month 1, medication and supply costs were significantly lower for surgery patients (P < 0.001). The therapeutic benefits of bariatric surgery on diabetes translate into considerable economic benefits. These data suggest that surgical therapy is clinically more effective and ultimately less expensive than standard therapy for diabetes patients with BMI ≥ 35 kg/m².

Germinal center function in the spleen during simian HIV infection in rhesus monkeys.

Infection with HIV-1, SIV, or simian HIV is associated with abnormalities in the number, size, and structure of germinal centers (GCs). To determine whether these histopathologic abnormalities are associated with abnormalities in Ab development, we analyzed nucleotide sequences of Igs from splenic GCs of simian HIV-infected macaques. Virus-specific GCs were identified in frozen splenic tissue sections by inverse immunohistochemistry using rHIV-1 gp120 as a probe. B cells from envelope-specific GCs were isolated from these sections using laser capture microdissection. Their Igs were amplified from cDNA using nested PCR, then cloned and sequenced. Nucleotide sequences were recovered from nine multimember clonal lineages. Within each lineage, sequences had similar V-D-J or V-J junctions but differed by somatic mutations distributed throughout the variable domain. The clones were highly mutated, similar to that previously reported for HIV-1-specific human IgG Abs. The average clone had 37 mutations in the V region, for a frequency of 0.11 mutations/base. The mutational pattern was strikingly nonrandom, with somatic mutations occurring preferentially at RGYW/WRCY hotspots. Transition mutations were favored over transversions, with C-->T and G-->A replacements together accounting for almost one-third of all mutations. Analysis of replacement and silent mutations in the framework and CDRs suggests that the Igs were subjected to affinity selection. These data demonstrate that the process of Ab maturation is not seriously disrupted in GCs during the early stages of immunodeficiency virus infection, and that Env-specific Igs developing in GCs are subject to extensive somatic mutation and profound selection pressures.

Expression of interleukin-18 and caspase-1 in cutaneous T-cell lymphoma.

PURPOSE: Cutaneous T-cell lymphoma (CTCL) is a malignancy of skin-homing Th2 T cells. Clonal T cells and CTCL skin lesions typically express Th2 cytokines, including interleukin (IL)-4, IL-5, and IL-10, but fail to produce Th1 cytokines. However, the reason for Th2 bias is unknown. IL-18 is a pleiotropic proinflammatory cytokine produced by monocytes/macrophages lineage as well as epithelial cells, such as human keratinocytes. In the absence of IL-12, IL-18 leads to increased immunoglobulin E production from B cells and enhanced production of IL-4 and IL-13 by basophils, mast cells, and CD4(+) T cells. We have analyzed cytokines in CTCL patients, which may bias the immune response around the Th1/Th2 axis. EXPERIMENTAL DESIGN: We examined plasma of 95 CTCL patients and skin of 20 CTCL patients for IL-18, caspase-1, IL-12, and other cytokines. To identify the presence or absence of these cytokine proteins in CTCL and normal skin, we cultured explants from skin biopsies on three-dimensional matrices. RESULTS: Plasma levels of IL-18 and its converting enzyme, caspase-1, were significantly elevated in CTCL. mRNA levels for these factors were also elevated in CTCL skin lesions. Matrices populated with CTCL lesional skin produced significant amounts of IL-18 and caspase-1; however, production of IL-12 protein was barely detectable. CONCLUSIONS: We propose that the high levels of IL-18 expression in lesional CTCL skin contribute to increased plasma levels of IL-18 and that this, in the face of significantly lower levels of IL-12, may contribute to the Th2 bias seen in this disease.

Skin-derived interleukin-7 contributes to the proliferation of lymphocytes in cutaneous T-cell lymphoma.

Cutaneous T-cell lymphomas (CTCLs) are malignancies of T cells that have a special affinity for the skin. We have previously reported that much of the T-cell receptor repertoire is altered in CTCL, and both malignant and nonmalignant clones are numerically expanded, presumably in response to T-cell trophic cytokines. We therefore examined levels of the T-cell trophic cytokines IL-2, IL-4, IL-7, IL-12, IL-13, and IL-15 in plasma in 93 CTCL patients and healthy controls. Only IL-7 levels were elevated in CTCL. We next looked at lesional skin from patients with CTCL and found elevated levels of IL-7 mRNA. Explant cultures of normal and lesional CTCL skin biopsies revealed significantly more IL-7 protein production in CTCL skin. Additionally, cultures of CTCL skin released greater numbers of T cells than normal skin; this was blocked by the addition of an IL-7 neutralizing antibody. Finally, these cultures induced proliferation of normal peripheral skin-homing T cells that were added to the cultures. These observations led us to postulate that IL-7 produced by skin cells contributes to the survival and proliferation of T cells within skin lesions and is likely the source of elevated circulating IL-7 in CTCL.

Immunization using autologous dendritic cells pulsed with the melanoma-associated antigen gp100-derived G280-9V peptide elicits CD8+ immunity.

PURPOSE: To determine the toxicity, maximal tolerated dose, and clinical and immunologic response to autologous dendritic cells pulsed with melanoma-associated antigen gp100-derived G280-9V peptide. PATIENTS AND METHODS: Twelve HLA-A*0201(+) patients with advanced melanoma were administered dendritic cells pulsed with G280-9V peptide. Cohorts of three patients were administered 5 x 10(6), 15 x 10(6), and 50 x 10(6) cells i.v. every 3 weeks for six doses according to a dose escalation scheme. Three additional patients were treated at the highest dose. No additional cytokines or therapies were coadministered. The immunogenicity of G280-9V-pulsed dendritic cells was measured by IFN-gamma ELISPOT assay, tetramer assay, and (51)Cr release assay comparing prevaccination to postvaccination blood samples. Response to treatment was assessed by Response Evaluation Criteria in Solid Tumors. RESULTS: CD8(+) immunity to the native G280 was observed in 8 (67%) patients as measured by ELISPOT and in 12 (100%) patients as measured by tetramer assay. Of the 9 patients tested, 9 (100%) had measurable high-avidity CTL activity as defined by lysis of allogeneic melanoma lines, which coexpress HLA-A*0201 and gp100. The median follow-up of the entire cohort is 43.8 months. Two (17%) partial responses were observed and 3 (25%) patients had stable disease. The median survival of the treated population was 37.6 months. At this time, three patients are alive, including one patient who continues to respond without additional treatment. CONCLUSION: The high rate of immunization as measured by three independent assays and the occurrence of clinical regression support continued investigation of G280-9V peptide as a candidate epitope in melanoma vaccine formulations.

Decreased T-cell receptor excision circles in cutaneous T-cell lymphoma.

PURPOSE: The T cell repertoire in patients with advanced cutaneous T cell lymphoma (CTCL) is significantly contracted despite the presence of relatively normal absolute numbers of T cells. We propose that many normal T cells were being lost in patients with CTCL, with the remaining normal T cells expanding clonally to fill the T cell compartment. T-cell receptor excision circles (TREC) form as a result of the initial gene rearrangement in naïve T cells. Although they are stable, they do not replicate and are subsequently diluted with the expansion of a population of T cells. Their concentration is therefore a measure of unexpanded naïve T cells relative to T cells that have undergone expansion. EXPERIMENTAL DESIGN: We analyzed TRECs from unfractionated peripheral blood T cells from 108 CTCL patients by quantitative PCR. In patients with obvious peripheral blood involvement, we also analyzed TRECs from clonal and nonclonal T cells. RESULTS: We found a decrease in the number of TRECs in peripheral blood of patients with CTCL at all stages of disease, and this decrease was proportional to the loss of complexity of the T cell repertoire as measured by complementarity-determining region 3 spectratyping. In patients with leukemic CTCL and a numerically expanded clone, we also found a significantly lower-than-expected number of TRECs in the nonclonal normal T cells. CONCLUSIONS: We hypothesize that the nonmalignant T cells have proliferated to fill the empty T cell repertoire space left by the loss of other T cells, leading to diminished TRECs and loss of T-cell receptor diversity.

Novel use of the BD FACS SPA to automate custom monoclonal antibody panel preparations for immunophenotyping.

BACKGROUND: The effective and accurate diagnosis of hematologic malignancies relies on flow cytometric immunophenotyping. Selected combinations of monoclonal antibodies (mAbs) arranged in multicolor panels allow for the accurate definition of normal and abnormal hematologic cell populations. The most time-consuming and crucial step in the staining process involves dispensing combinations of multiple mAbs into their appropriate staining tubes. This step is prone to error, requires concentration and accuracy, and is dependent on technologist experience. The Becton Dickinson BioScience (BD) FACS Sample Prep Assistant (SPA) is touted as a breakthrough in automated in vitro diagnostic sample preparation. The SPA is designed to automate BD MultiTESk and BD TriTest lyse/no-wash assays. However, because most cases in our laboratory require tedious application of unique four-color mAb cocktails for leukemia and lymphoma testing, we wondered whether the SPA would be helpful in accurately dispensing these mixtures. METHODS: The mAb panels were prepared by the SPA in two separate timed runs and on separate days. Eleven specimens (nine from patients and two from normal volunteers) were split and stained with four-color cocktails created by the SPA or manually. The percentage of positive (%P) cells and mean fluorescent intensity for each mAb pair were determined. These values were plotted against each other and correlation values were calculated. To quantitate timesaving in the laboratory, two technologists prepared individually the same mAb panels and were timed. RESULTS: The correlation between the two methods was high; r(2) was 0.988 for 158 %P antigen pairs; no bias between the manual and robotic methods was detected with the Wilcoxon rank test. Bland-Altman analysis indicated no obvious relation between the difference and the mean of %P cells, suggesting that the SPA successfully dispensed antibodies for leukemia/lymphoma panels. The two methods may be interchangeable, although the limited sample size prohibits this conclusion from Bland-Altman statistics alone. In addition, one possible error was detected in the SPA-prepared panels. The SPA averaged 65 min/run, the experienced technologist 12.95 min/run, and the inexperienced technologist 54.9 min/run. CONCLUSIONS: SPA dispensing time was twice the average manual dispensing time; however, SPA use was completely automated and freed the technologist to perform other tasks. SPA use permitted preemptive preparation of mAb panels and thus streamlined processing; however, the cost of the assay and the amount of reagent waste increased. It is certain that software modifications by BD could decrease the SPA reagent dispense time and decrease the cost associated with reagent waste when the SPA is used in this novel fashion.

Using local correlation in kernel-based smoothers for dependent data.

We consider the general problem of smoothing correlated data to estimate the nonparametric mean function when a random, but bounded, number of measurements is available for each independent subject. We propose a simple extension to the local polynomial regression smoother that retains the asymptotic properties of the working independence estimator, while typically reducing both the conditional bias and variance for practical sample sizes, as demonstrated by exact calculations for some particular models. We illustrate our method by smoothing longitudinal functional decline data for 100 patients with Huntington's disease. The class of local polynomial kernel-based estimating equations previously considered in the literature is shown to use the global correlation structure in an apparently detrimental way, which explains why some previous attempts to incorporate correlation were found to be asymptotically inferior to the working independence estimator.