Impact of interventions designed to increase market share and prescribing of fexofenadine at HMOs.

The impact of interventions designed to shift prescribing from loratadine to fexofenadine at HMOs was studied. Pharmacy claims data for a six-month preintervention period at four HMOs were analyzed to identify all new and refill prescriptions for loratadine, fexofenadine, astemizole, and cetirizine. The interventions consisted of a mandatory lockout of loratadine in favor of fexofenadine (at HMO A), a voluntary switch to fexofenadine promoted through letters to both physicians and members (HMO B), and a voluntary switch promoted through letters to physicians only (HMO C). There was no intervention at HMO D. Pharmacy claims data for the six months after each intervention program was implemented were analyzed to determine changes in the market share and prescribing of the study drugs. After the intervention programs were implemented, the market share of fexofenadine increased from 18.9% to 65.2% at HMO A, from 14.8% to 21.0% at HMO B, and from 20.7% to 23.8% at HMO C. Loratadine's market share decreased from 62.3% to 8.7% at HMO A, from 67.5% to 58.6% at HMO B, and from 70.5% to 65.3% at HMO C. HMOs A, B, and C each had greater shifts in market share for fexofenadine and loratadine than the control HMO. Changes in prescribing followed a similar pattern for the 25 physicians at each HMO who had most frequently prescribed loratadine during the preintervention period. The average cost per antihistamine prescription decreased 22.3% at HMO A. Prescription costs continued to rise at HMOs B, C, and D. Mandating the use of fexofenadine produced a significant increase in its market share, reduced the cost of nonsedating antihistamines, and successfully influenced prescribing behavior. Voluntary programs had a more modest impact on market share and did not stop increases in prescription costs.

HMO direct costs and health care resources use after implementation of a monthly limit on sumatriptan.

The health care costs and resource use of patients with migraine before and after a quantity limit on sumatriptan was introduced in an HMO were compared. A longitudinal, retrospective review of a medical claims database and a pharmacy claims database was conducted for two six-month periods before and after a monthly limit (four tablets or injections) on sumatriptan reimbursement was instituted at an independent practice association-model HMO in February 1997. Patients with at least one medical claim with a diagnosis code for migraine or at least two pharmacy claims for sumatriptan, methysergide, ergotamine, dihydroergotamine, or an ergotamine combination product in 1996 or 1997 were eligible for inclusion. A total of 557 patients were included in the analysis. Migraine-related medical costs and total medical costs increased 1.5% and 24.4%, respectively; neither change was statistically significant. Physician office visits related to migraine increased by 7.8%. The number of hospital admissions for the cohort increased from three to five, but hospital costs decreased by 55.0%. The overall costs of medications for migraine therapy decreased by 4.5%. There was an 8.2% increase in prescriptions for drugs to treat migraine but a 40.0% decrease in their cost, primarily because of decreased sumatriptan use. There was a 33.9% increase in prescriptions for medications that could be used as prophylaxis for migraine and a 49.6% increase in their cost. Implementation of a monthly limit on sumatriptan decreased an HMO's pharmacy costs but did not significantly alter migraine-related direct medical costs and health care resource use of patients with migraine.

Direct health care costs for treatment of diabetes mellitus and hypertension in an IPA-group-model HMO.

The differences in direct health care costs and use between HMO enrollees with both diabetes mellitus and hypertension and enrollees with either disease alone were studied. Two years' worth of medical and pharmacy claims data from a hybrid (independent practice association and group)-model HMO were evaluated. Diagnoses were determined from medical claims data and cross-referenced with prescription information from pharmacy claims data. Aggregate costs associated with each disease, including pharmacy costs, costs of physician office visits, and laboratory costs, were compiled. Comparisons were made of all costs (any cost incurred by the health plan for the member, regardless of disease) and disease-specific costs. The frequency of comorbid conditions was identified. A total of 6195 patients (670 with diabetes and hypertension, 1756 with diabetes alone, and 3769 with hypertension alone) were assessed. Patients with both diseases incurred much higher costs per year than patients with diabetes or hypertension alone (mean costs, $13,446, $8,493, and $8,424, respectively). Hospitalization costs contributed the greatest amount to total costs, while emergency room costs contributed the least. Disease-specific costs for diabetes and hypertension represented less than one quarter of total health care costs per patient. Average disease-specific costs were highest for patients with both diseases ($2,955), followed by costs for patients with hypertension alone ($1,803) and patients with diabetes alone ($689). The percentage spent on prescriptions was much higher for disease-specific costs than for total costs. The three most common comorbid conditions were dyslipidemia, coronary artery disease, and chronic obstructive pulmonary disease, with the frequency of cerebrovascular disease and myocardial infarction more than double in patients with diabetes and hypertension compared with patients with either disease alone. The cost of care for a patient with both diabetes and hypertension, although not double that for a patient with diabetes or hypertension alone, was higher than the cost of treating either disease.

Integrating computation and visualization for biomolecular analysis: an example using python and AVS.

One of the challenges in biocomputing is to enable the efficient use of a wide variety of fast-evolving computational methods to simulate, analyze, and understand the complex properties and interactions of molecular systems. Our laboratory investigates several areas including molecular visualization, protein-ligand docking, protein-protein docking, molecular surfaces, and the derivation of phenomenological potentials. In this paper we present an approach based on the Python programming language to achieve a high level of integration between these different computational methods and our primary visualization system AVS. This approach removes many limitations of AVS while increasing dramatically the inter-operability of our computational tools. Several examples are shown to illustrate how this approach enables a high level of integration and inter-operability between different tools, while retaining modularity and avoiding the creation of a large monolithic package that is difficult to extend and maintain.

Interactive modeling of supramolecular assemblies.

The modeling of supramolecular structure presents two major challenges: (1) managing the large amount of sequence, structural and biochemical data, and (2) presenting the data to the user in a flexible and comprehensible manner that addresses these problems. We describe a visualization environment for the creation and analysis of supramolecular models. A set of modular symmetry tools, collectively called SymGen, has been created, providing a flexible platform for the creation of complex assemblies, with interactive control of all symmetry elements and their parameters. A second tool, SymSearch, allows a range of parameters defined within SymGen to be sampled and the resulting conformations to be evaluated. The environment avoids information overload, caused by the large number of atoms in supramolecular complexes, by using a multiresolution spherical harmonic representation that allows the user to display only essential features. Spherical harmonics also enables control of the triangulation level, allowing the user to reduce the complexity of the geometric description to retain interactive speed. The visual fidelity of the surface data is retained by using texture maps that are independent of the resolution of the underlying triangulation. We describe the design and implementation of this environment, and three illustrative examples of its utility.

Biomolecular visualization using AVS.

Dataflow systems for scientific visualization are becoming increasingly sophisticated in their architecture and functionality. AVS, from Advanced Visual Systems Inc., is a powerful dataflow environment that has been applied to many computation and visualization tasks. An important, yet complex, application area is molecular modeling and biomolecular visualization. Problems in biomolecular visualization tax the capability of dataflow systems because of the diversity of operations that are required and because many operations do not fit neatly into the dataflow paradigm. Here we describe visualization strategies and auxiliary programs developed to enhance the applicability of AVS for molecular modelling. Our visualization strategy is to use general-purpose AVS modules and a small number of chemistry-specific modules. We have developed methods to control AVS using AVS-tool, a programmable interface to the AVS Command Line Interpreter (CLI), and have also developed NAB, a C-like language for writing AVS modules that has extensions for operating on proteins and nucleic acids. This strategy provides a flexible and extensible framework for a wide variety of molecular modeling tasks.

Approximation and visualization of large-scale motion of protein surfaces.

We present a method for the approximation and real-time visualization of large-scale motion of protein surfaces. A molecular surface is represented by an expansion of spherical harmonic functions, and the motion of protein atoms around their equilibrium positions is computed by normal mode analysis. The motion of the surface is approximated by projecting the normal mode vectors of the solvent-accessible atoms to the spherical harmonic representation of the molecular surface. These surface motion vectors are represented by a separate spherical harmonic expansion. Representing the surface geometry and the surface motion vectors by spherical harmonic expansions allows variable-resolution analysis and real-time display of the large-scale surface motion. This technique has been applied to interactive visualization, interactive surface manipulation, and animation.

Texture mapping parametric molecular surfaces.

Texture mapping is an increasingly popular technique in molecular modeling. It is particularly effective in representing high-resolution surface detail using a low-resolution polygonal model. We describe how texture mapping can be used with parametric molecular surfaces represented as expansions of spherical harmonic functions. We define analytically the texture image and its transformation to a parametric surface. Unlike most methods of texture mapping, this transformation defines a one-to-one correspondence between the surface and the texture; texture coordinates are derived from the location of the surface point and not from physical properties at the surface point. This has advantages for the interactive visualization of surface data. We control the interactive response time by lowering the resolution of the polygon mesh while retaining the high-resolution detail of the texture, or we can lower the resolution of the texture image with the same polygonal model. By using a well-defined convention for texture coordinates, we can use the same image for the original surface or its parametric representation, and we can rapidly switch between images that represent different surface properties without recomputing the texture coordinates. Parametric surfaces allow new flexibility for the visualization of molecular surface data.

Approximation and characterization of molecular surfaces.

The representation and characterization of molecular surfaces are important in many areas of molecular modeling. Parametric representations of protein molecular surfaces are a compact way to describe a surface, and are useful for the evaluation of surface properties such as the normal vector, principal curvatures, and principal curvature directions. Simplified representations of molecular surfaces are useful for efficient rendering and for the display of large-scale surface features. Several techniques for representing surfaces by expansions of spherical harmonic functions have been reported, but these techniques require that the radius function is single valued, that is, each ray from an origin inside the surface intersects the surface at one and only one point. A new technique is described that removes this limitation and can be used to compute surface shape properties.

Shape analysis of molecular surfaces.

The description of molecular shape is important in the analysis of protein-protein and protein-ligand interactions. We describe volumetric and surface-based techniques for computing shape properties of molecular surfaces. The surface is defined as an isocontour of an approximate electron density function. Each technique can compute several scalar and vector surface properties such as the Gaussian and mean curvature, principal curvatures, and principal curvature directions. Shape properties are derived from the eigenvalues and eigenvectors of a 3 by 3 matrix for each surface point. In the volumetric approach, the matrix is the second derivative of an approximate electron density function. In the surface-based approach, the matrix is the approximate gradient of the surface normal. Derivatives are computed by convolving the density or the surface normals with the derivatives of a Gaussian function. The variance of the Gaussian determines the effective length scale at which the surface is analyzed. Scalar surface properties are displayed as colored dots or shaded triangles, and vector properties are displayed as line segments from each surface point. This report describes the implementation of these procedures and their use in computing the shape properties of Cu-Zu superoxide dismutase.

Heuristic refinement method for the derivation of protein solution structures: validation on cytochrome b562.

A method is described for determining the family of protein structures compatible with solution data obtained primarily from nuclear magnetic resonance (NMR) spectroscopy. Starting with all possible conformations, the method systematically excludes conformations until the remaining structures are only those compatible with the data. The apparent computational intractability of this approach is reduced by assembling the protein in pieces, by considering the protein at several levels of abstraction, by utilizing constraint satisfaction methods to consider only a few atoms at a time, and by utilizing artificial intelligence methods of heuristic control to decide which actions will exclude the most conformations. Example results are presented for simulated NMR data from the known crystal structure of cytochrome b562 (103 residues). For 10 sample backbones an average root-mean-square deviation from the crystal of 4.1 A was found for all alpha-carbon atoms and 2.8 A for helix alpha-carbons alone. The 10 backbones define the family of all structures compatible with the data and provide nearly correct starting structures for adjustment by any of the current structure determination methods.