Multicenter epidemiologic and health services research on therapeutics in the HMO Research Network Center for Education and Research on Therapeutics

A tree-based scan statistic for zero-inflated count data in post-market drug safety surveillance

After new drugs enter the market, adverse events (AE) induced by their use must be tracked; rare AEs may not be detected during clinical trials. Some organizations have been collecting information on suspected drugs and AEs via a spontaneous reporting system to conduct post-market drug safety surveillance. These organizations use the information to detect a signal representing potential causality between drugs and AEs. The drug and AE data are often hierarchically structured. Accordingly, the tree-based scan statistic can be used as a statistical data mining method for signal detection. Most of the AE databases contain a large number of zero-count cells. Notably, not only an observational zero from the Poisson distribution, but also a true zero exists in zero-count cells. True zeros represent theoretically impossible observations or possible but unreported observations. The existing tree-based scan statistic assumes that all zeros are zero-valued observations from the Poisson distribution. Therefore, true zeros are not considered in the modeling, which can lead to bias in the inferences. In this study, we propose a tree-based scan statistic for zero-inflated count data in a hierarchical structure. According to our simulation study, in the presence of excess zeros, our proposed tree-based scan statistic provides better performance than the existing tree-based scan statistic. The two methods were illustrated using Korea Adverse Event Reporting System data from the Korea Institute of Drug Safety and Risk Management.

A Governance Framework to Integrate Longitudinal Clinical and Community Data in a Distributed Data Network: The Childhood Obesity Data Initiative

CONTEXT

Integrating longitudinal data from community-based organizations (eg, physical activity programs) with electronic health record information can improve capacity for childhood obesity research.

OBJECTIVE

A governance framework that protects individual privacy, accommodates organizational data stewardship requirements, and complies with laws and regulations was developed and implemented to support the harmonization of data from disparate clinical and community information systems.

PARTICIPANTS AND SETTING

Through the Childhood Obesity Data Initiative (CODI), 5 Colorado-based organizations collaborated to expand an existing distributed health data network (DHDN) to include community-generated data and assemble longitudinal patient records for research.

DESIGN

A governance work group expanded an existing DHDN governance infrastructure with CODI-specific data use and exchange policies and procedures that were codified in a governance plan and a delegated-authority, multiparty, reciprocal agreement.

RESULTS

A CODI governance work group met from January 2019 to March 2020 to conceive an approach, develop documentation, and coordinate activities. Governance requirements were synthesized from the CODI use case, and a customized governance approach was constructed to address governance gaps in record linkage, a procedure to request data, and harmonizing community and clinical data. A Master Sharing and Use Agreement (MSUA) and Memorandum of Understanding were drafted and executed to support creation of linked longitudinal records of clinical- and community-derived childhood obesity data. Furthermore, a multiparty infrastructure protocol was approved by the local institutional review board (IRB) to expedite future CODI research by simplifying IRB research applications.

CONCLUSION

CODI implemented a clinical-community governance strategy that built trust between organizations and allowed efficient data exchange within a DHDN. A thorough discovery process allowed CODI stakeholders to assess governance capacity and reveal regulatory and organizational obstacles so that the governance infrastructure could effectively leverage existing knowledge and address challenges. The MSUA and complementary governance documents can inform similar efforts.

Trends in antifungal use in US hospitals, 2006-12

Background

Although trends in antibiotic use have been characterized, less is known about antifungal use. Data on antifungal use are important for understanding practice patterns, assessing emergence of antifungal resistance and developing antifungal stewardship programmes. We estimated national trends in inpatient antifungal use in the USA.

Methods

Using billing data for antifungals from the Truven Health MarketScan® Hospital Drug Database during 2006-12, we estimated the proportion of discharges at which antifungals were given and days of therapy (DOT)/1000 patient days (PDs) by antifungal drug type, year, patient and facility characteristics. We created national estimates using weights generated from Centers for Medicare and Medicaid Services data and assessed trends over time.

Results

Overall, 2.7% of all inpatients and 7.7% of those in ICUs received antifungals. The estimated DOT/1000 PDs for any antifungal was 35.0 for all inpatients and 73.7 for ICU patients. Azoles accounted for 80% of all antifungal use (28.5/1000 PDs), followed by echinocandins (5.0/1000 PDs). By multivariable trend analysis, DOT/1000 PDs for azoles (-21%) and polyenes (-47%) decreased between 2006 and 2012, whereas echinocandins increased 11% during 2006-10 and declined after 2011. Unspecified septicaemia, HIV and antineoplastic therapy were among the top primary diagnosis codes for patients who received antifungals.

Conclusions

Antifungals were most frequently used in ICU settings and fluconazole accounted for a large, but declining, proportion of antifungal use. Antifungal stewardship efforts may have the most impact if focused in ICUs, among certain patient groups (e.g. HIV and malignancy) and on stopping empirical antifungal therapy for unspecified sepsis when not indicated.

Improving pragmatic clinical trial design using real-world data

BACKGROUND

Pragmatic clinical trials often use automated data sources such as electronic health records, claims, or registries to identify eligible individuals and collect outcome information. A specific advantage that this automated data collection often yields is having data on potential participants when design decisions are being made. We outline how this data can be used to inform trial design.

METHODS

Our work is motivated by a pragmatic clinical trial evaluating the impact of suicide-prevention outreach interventions on fatal and non-fatal suicide attempts in the 18 months after randomization. We illustrate our recommended approaches for designing pragmatic clinical trials using historical data from the health systems participating in this study. Specifically, we illustrate how electronic health record data can be used to inform the selection of trial eligibility requirements, to estimate the distribution of participant characteristics over the course of the trial, and to conduct power and sample size calculations.

RESULTS

Data from 122,873 people with patient health questionnaire (PHQ) responses, recorded in their electronic health records between 1 July 2010 and 31 March 2012, were used to show that the suicide attempt rate in the 18 months following completion of the questionnaire varies by response to item nine of the PHQ. We estimated that the proportion of individuals with a prior recorded elevated PHQ (i.e. history of suicidal ideation) would decrease from approximately 50% at the beginning of a trial to about 5%, 50 weeks later. Using electronic health record data, we conducted simulations to estimate the power to detect a 25% reduction in suicide attempts. Simulation-based power calculations estimated that randomizing 8000 participants per randomization arm would allow 90% power to detect a 25% reduction in the suicide attempt rate in the intervention arm compared to usual care at an alpha rate of 0.05.

CONCLUSIONS

Historical data can be used to inform the design of pragmatic clinical trials, a strength of trials that use automated data collection for randomizing participants and assessing outcomes. In particular, realistic sample size calculations can be conducted using real-world data from the health systems in which the trial will be conducted. Data-informed trial design should yield more realistic estimates of statistical power and maximize efficiency of trial recruitment.

Lessons from meta-analyses of randomized clinical trials for analysis of distributed networks of observational databases

Networks of constellations of longitudinal observational databases, often electronic medical records or transactional insurance claims or both, are increasingly being used for studying the effects of medicinal products in real-world use. Such databases are frequently configured as distributed networks. That is, patient-level data are kept behind firewalls and not communicated outside of the data vendor other than in aggregate form. Instead, data are standardized across the network, and queries of the network are executed locally by data partners, and summary results provided to a central research partner(s) for amalgamation, aggregation, and summarization. Such networks can be huge covering years of data on upwards of 100 million patients. Examples of such networks include the FDA Sentinel Network, ASPEN, CNODES, and EU-ADR. As this is a new emerging field, we note in this paper the conceptual similarities and differences between the analysis of distributed networks and the now well-established field of meta-analysis of randomized clinical trials (RCTs). We recommend, wherever appropriate, to apply learnings from meta-analysis to help guide the development of distributed network analyses of longitudinal observational databases.

Leveraging a Statewide Clinical Data Warehouse to Expand Boundaries of the Learning Health System

Learning Health Systems (LHS) require accessible, usable health data and a culture of collaboration—a challenge for any single system, let alone disparate organizations, with macro- and micro-systems. Recently, the National Science Foundation described this important setting as a cyber-social ecosystem. In 2004, in an effort to create a platform for transforming health in South Carolina, Health Sciences South Carolina (HSSC) was established as a research collaboration of the largest health systems, academic medical centers and research intensive universities in South Carolina. With work beginning in 2010, HSSC unveiled an integrated Clinical Data Warehouse (CDW) in 2013 as a crucial anchor to a statewide LHS. This CDW integrates data from independent health systems in near-real time, and harmonizes the data for aggregation and use in research. With records from over 2.7 million unique patients spanning 9 years, this multi-institutional statewide clinical research repository allows integrated individualized patient-level data to be used for multiple population health and biomedical research purposes. In the first 21 months of operation, more than 2,800 de-identified queries occurred through i2b2, with 116 users. HSSC has developed and implemented solutions to complex issues emphasizing anti-competitiveness and participatory governance, and serves as a recognized model to organizations working to improve healthcare quality by extending the traditional borders of learning health systems.

Developing an Algorithm to Identify History of Cancer Using Electronic Medical Records

Introduction/Objective:

The objective of this study was to develop an algorithm to identify Kaiser Permanente Colorado (KPCO) members with a history of cancer.

Background:

Tumor registries are used with high precision to identify incident cancer, but are not designed to capture prevalent cancer within a population. We sought to identify a cohort of adults with no history of cancer, and thus, we could not rely solely on the tumor registry.

Methods:

We included all KPCO members between the ages of 40–75 years who were continuously enrolled during 2013 (N=201,787). Data from the tumor registry, chemotherapy files, inpatient and outpatient claims were used to create an algorithm to identify members with a high likelihood of cancer. We validated the algorithm using chart review and calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for occurrence of cancer.

Findings:

The final version of the algorithm achieved a sensitivity of 100 percent and specificity of 84.6 percent for identifying cancer. If we relied on the tumor registry alone, 47 percent of those with a history of cancer would have been missed.

Discussion:

Using the tumor registry alone to identify a cohort of patients with prior cancer is not sufficient. In the final version of the algorithm, the sensitivity and PPV were improved when a diagnosis code for cancer was required to accompany oncology visits or chemotherapy administration.

Conclusion:

Electronic medical record (EMR) data can be used effectively in combination with data from the tumor registry to identify health plan members with a history of cancer.

Preparing for the ICD-10-CM Transition: Automated Methods for Translating ICD Codes in Clinical Phenotype Definitions

Background:

The national mandate for health systems to transition from ICD-9-CM to ICD-10-CM in October 2015 has an impact on research activities. Clinical phenotypes defined by ICD-9-CM codes need to be converted to ICD-10-CM, which has nearly four times more codes and a very different structure than ICD-9-CM.

Methods:

We used the Centers for Medicare & Medicaid Services (CMS) General Equivalent Maps (GEMs) to translate, using four different methods, condition-specific ICD-9-CM code sets used for pragmatic trials (n=32) into ICD-10-CM. We calculated the recall, precision, and F score of each method. We also used the ICD-9-CM and ICD-10-CM value sets defined for electronic quality measure as an additional evaluation of the mapping methods.

Results:

The forward-backward mapping (FBM) method had higher precision, recall and F-score metrics than simple forward mapping (SFM). The more aggressive secondary (SM) and tertiary mapping (TM) methods resulted in higher recall but lower precision. For clinical phenotype definition, FBM was the best (F=0.67), but was close to SM (F=0.62) and TM (F=0.60), judging on the F-scores alone. The overall difference between the four methods was statistically significant (one-way ANOVA, F=5.749, p=0.001). However, pairwise comparisons between FBM, SM, and TM did not reach statistical significance. A similar trend was found for the quality measure value sets.

Discussion:

The optimal method for using the GEMs depends on the relative importance of recall versus precision for a given use case. It appears that for clinically distinct and homogenous conditions, the recall of FBM is sufficient. The performance of all mapping methods was lower for heterogeneous conditions. Since code sets used for phenotype definition and quality measurement can be very similar, there is a possibility of cross-fertilization between the two activities.

Conclusion:

Different mapping approaches yield different collections of ICD-10-CM codes. All methods require some level of human validation.

A Naive Bayes machine learning approach to risk prediction using censored, time-to-event data

Predicting an individual's risk of experiencing a future clinical outcome is a statistical task with important consequences for both practicing clinicians and public health experts. Modern observational databases such as electronic health records provide an alternative to the longitudinal cohort studies traditionally used to construct risk models, bringing with them both opportunities and challenges. Large sample sizes and detailed covariate histories enable the use of sophisticated machine learning techniques to uncover complex associations and interactions, but observational databases are often 'messy', with high levels of missing data and incomplete patient follow-up. In this paper, we propose an adaptation of the well-known Naive Bayes machine learning approach to time-to-event outcomes subject to censoring. We compare the predictive performance of our method with the Cox proportional hazards model which is commonly used for risk prediction in healthcare populations, and illustrate its application to prediction of cardiovascular risk using an electronic health record dataset from a large Midwest integrated healthcare system.

Implementation of new clinical programs in the VHA healthcare system: the importance of early collaboration between clinical leadership and research

Collaboration between policy, research, and clinical partners is crucial to achieving proven quality care. The Veterans Health Administration has expended great efforts towards fostering such collaborations. Through this, we have learned that an ideal collaboration involves partnership from the very beginning of a new clinical program, so that the program is designed in a way that ensures quality, validity, and puts into place the infrastructure necessary for a reliable evaluation. This paper will give an example of one such project, the Lung Cancer Screening Demonstration Project (LCSDP). We will outline the ways that clinical, policy, and research partners collaborated in design, planning, and implementation in order to create a sustainable model that could be rigorously evaluated for efficacy and fidelity. We will describe the use of the Donabedian quality matrix to determine the necessary characteristics of a quality program and the importance of the linkage with engineering, information technology, and clinical paradigms to connect the development of an on-the-ground clinical program with the evaluation goal of a learning healthcare organization. While the LCSDP is the example given here, these partnerships and suggestions are salient to any healthcare organization seeking to implement new scientifically proven care in a useful and reliable way.

Developing a data infrastructure for a learning health system: the PORTAL network

The Kaiser Permanente & Strategic Partners Patient Outcomes Research To Advance Learning (PORTAL) network engages four healthcare delivery systems (Kaiser Permanente, Group Health Cooperative, HealthPartners, and Denver Health) and their affiliated research centers to create a new national network infrastructure that builds on existing relationships among these institutions. PORTAL is enhancing its current capabilities by expanding the scope of the common data model, paying particular attention to incorporating patient-reported data more systematically, implementing new multi-site data governance procedures, and integrating the PCORnet PopMedNet platform across our research centers. PORTAL is partnering with clinical research and patient experts to create cohorts of patients with a common diagnosis (colorectal cancer), a rare diagnosis (adolescents and adults with severe congenital heart disease), and adults who are overweight or obese, including those with pre-diabetes or diabetes, to conduct large-scale observational comparative effectiveness research and pragmatic clinical trials across diverse clinical care settings.

Clinical research data warehouse governance for distributed research networks in the USA: a systematic review of the literature

OBJECTIVE

To review the published, peer-reviewed literature on clinical research data warehouse governance in distributed research networks (DRNs).

MATERIALS AND METHODS

Medline, PubMed, EMBASE, CINAHL, and INSPEC were searched for relevant documents published through July 31, 2013 using a systematic approach. Only documents relating to DRNs in the USA were included. Documents were analyzed using a classification framework consisting of 10 facets to identify themes.

RESULTS

6641 documents were retrieved. After screening for duplicates and relevance, 38 were included in the final review. A peer-reviewed literature on data warehouse governance is emerging, but is still sparse. Peer-reviewed publications on UK research network governance were more prevalent, although not reviewed for this analysis. All 10 classification facets were used, with some documents falling into two or more classifications. No document addressed costs associated with governance.

DISCUSSION

Even though DRNs are emerging as vehicles for research and public health surveillance, understanding of DRN data governance policies and procedures is limited. This is expected to change as more DRN projects disseminate their governance approaches as publicly available toolkits and peer-reviewed publications.

CONCLUSIONS

While peer-reviewed, US-based DRN data warehouse governance publications have increased, DRN developers and administrators are encouraged to publish information about these programs.

Clinical research data warehouse governance for distributed research networks in the USA: a systematic review of the literature

OBJECTIVE

To review the published, peer-reviewed literature on clinical research data warehouse governance in distributed research networks (DRNs).

MATERIALS AND METHODS

Medline, PubMed, EMBASE, CINAHL, and INSPEC were searched for relevant documents published through July 31, 2013 using a systematic approach. Only documents relating to DRNs in the USA were included. Documents were analyzed using a classification framework consisting of 10 facets to identify themes.

RESULTS

6641 documents were retrieved. After screening for duplicates and relevance, 38 were included in the final review. A peer-reviewed literature on data warehouse governance is emerging, but is still sparse. Peer-reviewed publications on UK research network governance were more prevalent, although not reviewed for this analysis. All 10 classification facets were used, with some documents falling into two or more classifications. No document addressed costs associated with governance.

DISCUSSION

Even though DRNs are emerging as vehicles for research and public health surveillance, understanding of DRN data governance policies and procedures is limited. This is expected to change as more DRN projects disseminate their governance approaches as publicly available toolkits and peer-reviewed publications.

CONCLUSIONS

While peer-reviewed, US-based DRN data warehouse governance publications have increased, DRN developers and administrators are encouraged to publish information about these programs.

The HMO Research Network Virtual Data Warehouse: A Public Data Model to Support Collaboration

The HMO Research Network (HMORN) Virtual Data Warehouse (VDW) is a public, non-proprietary, research-focused data model implemented at 17 health care systems across the United States. The HMORN has created a governance structure and specified policies concerning the VDW's content, development, implementation, and quality assurance. Data extracted from the VDW have been used by thousands of studies published in peer-reviewed journal articles. Advances in software supporting care delivery and claims processing and the availability of new data sources have greatly expanded the data available for research, but substantially increased the complexity of data management. The VDW data model incorporates software and data advances to ensure that comprehensive, up-to-date data of known quality are available for research. VDW governance works to accommodate new data and system complexities. This article highlights the HMORN VDW data model, its governance principles, data content, and quality assurance procedures. Our goal is to share the VDW data model and its operations to those wishing to implement a distributed interoperable health care data system.

Data quality assessment for comparative effectiveness research in distributed data networks

BACKGROUND

Electronic health information routinely collected during health care delivery and reimbursement can help address the need for evidence about the real-world effectiveness, safety, and quality of medical care. Often, distributed networks that combine information from multiple sources are needed to generate this real-world evidence.

OBJECTIVE

We provide a set of field-tested best practices and a set of recommendations for data quality checking for comparative effectiveness research (CER) in distributed data networks.

METHODS

Explore the requirements for data quality checking and describe data quality approaches undertaken by several existing multi-site networks.

RESULTS

There are no established standards regarding how to evaluate the quality of electronic health data for CER within distributed networks. Data checks of increasing complexity are often used, ranging from consistency with syntactic rules to evaluation of semantics and consistency within and across sites. Temporal trends within and across sites are widely used, as are checks of each data refresh or update. Rates of specific events and exposures by age group, sex, and month are also common.

DISCUSSION

Secondary use of electronic health data for CER holds promise but is complex, especially in distributed data networks that incorporate periodic data refreshes. The viability of a learning health system is dependent on a robust understanding of the quality, validity, and optimal secondary uses of routinely collected electronic health data within distributed health data networks. Robust data quality checking can strengthen confidence in findings based on distributed data network.

Drug Adverse Event Detection in Health Plan Data Using the Gamma Poisson Shrinker and Comparison to the Tree-based Scan Statistic

Background: Drug adverse event (AE) signal detection using the Gamma Poisson Shrinker (GPS) is commonly applied in spontaneous reporting. AE signal detection using large observational health plan databases can expand medication safety surveillance. Methods: Using data from nine health plans, we conducted a pilot study to evaluate the implementation and findings of the GPS approach for two antifungal drugs, terbinafine and itraconazole, and two diabetes drugs, pioglitazone and rosiglitazone. We evaluated 1676 diagnosis codes grouped into 183 different clinical concepts and four levels of granularity. Several signaling thresholds were assessed. GPS results were compared to findings from a companion study using the identical analytic dataset but an alternative statistical method—the tree-based scan statistic (TreeScan). Results: We identified 71 statistical signals across two signaling thresholds and two methods, including closely-related signals of overlapping diagnosis definitions. Initial review found that most signals represented known adverse drug reactions or confounding. About 31% of signals met the highest signaling threshold. Conclusions: The GPS method was successfully applied to observational health plan data in a distributed data environment as a drug safety data mining method. There was substantial concordance between the GPS and TreeScan approaches. Key method implementation decisions relate to defining exposures and outcomes and informed choice of signaling thresholds.

Building the informatics infrastructure for comparative effectiveness research (CER): a review of the literature

BACKGROUND

Technological advances in clinical informatics have made large amounts of data accessible and potentially useful for research. As a result, a burgeoning literature addresses efforts to bridge the fields of health services research and biomedical informatics. The Electronic Data Methods Forum review examines peer-reviewed literature at the intersection of comparative effectiveness research and clinical informatics. The authors are specifically interested in characterizing this literature and identifying cross-cutting themes and gaps in the literature.

METHODS

A 3-step systematic literature search was conducted, including a structured search of PubMed, manual reviews of articles from selected publication lists, and manual reviews of research activities based on prospective electronic clinical data. Two thousand four hundred thirty-five citations were identified as potentially relevant. Ultimately, a full-text review was performed for 147 peer-reviewed papers.

RESULTS

One hundred thirty-two articles were selected for inclusion in the review. Of these, 88 articles are the focus of the discussion in this paper. Three types of articles were identified, including papers that: (1) provide historical context or frameworks for using clinical informatics for research, (2) describe platforms and projects, and (3) discuss issues, challenges, and applications of natural language processing. In addition, 2 cross-cutting themes emerged: the challenges of conducting research in the absence of standardized ontologies and data collection; and unique data governance concerns related to the transfer, storage, deidentification, and access to electronic clinical data. Finally, the authors identified several current gaps on important topics such as the use of clinical informatics for cohort identification, cloud computing, and single point access to research data.

A pragmatic framework for single-site and multisite data quality assessment in electronic health record-based clinical research

INTRODUCTION

Answers to clinical and public health research questions increasingly require aggregated data from multiple sites. Data from electronic health records and other clinical sources are useful for such studies, but require stringent quality assessment. Data quality assessment is particularly important in multisite studies to distinguish true variations in care from data quality problems.

METHODS

We propose a "fit-for-use" conceptual model for data quality assessment and a process model for planning and conducting single-site and multisite data quality assessments. These approaches are illustrated using examples from prior multisite studies.

APPROACH

Critical components of multisite data quality assessment include: thoughtful prioritization of variables and data quality dimensions for assessment; development and use of standardized approaches to data quality assessment that can improve data utility over time; iterative cycles of assessment within and between sites; targeting assessment toward data domains known to be vulnerable to quality problems; and detailed documentation of the rationale and outcomes of data quality assessments to inform data users. The assessment process requires constant communication between site-level data providers, data coordinating centers, and principal investigators.

DISCUSSION

A conceptually based and systematically executed approach to data quality assessment is essential to achieve the potential of the electronic revolution in health care. High-quality data allow "learning health care organizations" to analyze and act on their own information, to compare their outcomes to peers, and to address critical scientific questions from the population perspective.

Electronic medical records as a tool in clinical pharmacology: opportunities and challenges

The development and increasing sophistication of electronic medical record (EMR) systems hold the promise of not only improving patient care but also providing unprecedented opportunities for discovery in the fields of basic, translational, and implementation sciences. Clinical pharmacology research in the EMR environment has only recently started to become a reality, with EMRs becoming increasingly populated, methods to mine drug response and other phenotypes becoming more sophisticated, and links being established with DNA repositories.

A secure distributed logistic regression protocol for the detection of rare adverse drug events

Background

There is limited capacity to assess the comparative risks of medications after they enter the market. For rare adverse events, the pooling of data from multiple sources is necessary to have the power and sufficient population heterogeneity to detect differences in safety and effectiveness in genetic, ethnic and clinically defined subpopulations. However, combining datasets from different data custodians or jurisdictions to perform an analysis on the pooled data creates significant privacy concerns that would need to be addressed. Existing protocols for addressing these concerns can result in reduced analysis accuracy and can allow sensitive information to leak.

Objective

To develop a secure distributed multi-party computation protocol for logistic regression that provides strong privacy guarantees.

Methods

We developed a secure distributed logistic regression protocol using a single analysis center with multiple sites providing data. A theoretical security analysis demonstrates that the protocol is robust to plausible collusion attacks and does not allow the parties to gain new information from the data that are exchanged among them. The computational performance and accuracy of the protocol were evaluated on simulated datasets.

Results

The computational performance scales linearly as the dataset sizes increase. The addition of sites results in an exponential growth in computation time. However, for up to five sites, the time is still short and would not affect practical applications. The model parameters are the same as the results on pooled raw data analyzed in SAS, demonstrating high model accuracy.

Conclusion

The proposed protocol and prototype system would allow the development of logistic regression models in a secure manner without requiring the sharing of personal health information. This can alleviate one of the key barriers to the establishment of large-scale post-marketing surveillance programs. We extended the secure protocol to account for correlations among patients within sites through generalized estimating equations, and to accommodate other link functions by extending it to generalized linear models.

Comparative-effectiveness research in distributed health data networks

Comparative-effectiveness research (CER) can be conducted within a distributed health data network. Such networks allow secure access to separate data sets from different data partners and overcome many practical obstacles related to patient privacy, data security, and proprietary concerns. A scalable network architecture supports a wide range of CER activities and meets the data infrastructure needs envisioned by the Federal Coordinating Council for Comparative Effectiveness Research.

A common 5'-UTR variant in MATE2-K is associated with poor response to metformin

Multidrug and toxin extrusion 2 (MATE2-K (SLC47A2)), a polyspecific organic cation exporter, facilitates the renal elimination of the antidiabetes drug metformin. In this study, we characterized genetic variants of MATE2-K, determined their association with metformin response, and elucidated their impact by means of a comparative protein structure model. Four nonsynonymous variants and four variants in the MATE2-K basal promoter region were identified from ethnically diverse populations. Two nonsynonymous variants-c.485C>T and c.1177G>A-were shown to be associated with significantly lower metformin uptake and reduction in protein expression levels. MATE2-K basal promoter haplotypes containing the most common variant, g.-130G>A (>26% allele frequency), were associated with a significant increase in luciferase activities and reduced binding to the transcriptional repressor myeloid zinc finger 1 (MZF-1). Patients with diabetes who were homozygous for g.-130A had a significantly poorer response to metformin treatment, assessed as relative change in glycated hemoglobin (HbA1c) (-0.027 (-0.076, 0.033)), as compared with carriers of the reference allele, g.-130G (-0.15 (-0.17, -0.13)) (P=0.002). Our study showed that MATE2-K plays a role in the antidiabetes response to metformin.

Improving automated case finding for ectopic pregnancy using a classification algorithm

BACKGROUND

Research and surveillance work addressing ectopic pregnancy often rely on diagnosis and procedure codes available from automated data sources. However, the use of these codes may result in misclassification of cases. Our aims were to evaluate the accuracy of standard ectopic pregnancy codes; and, through the use of additional automated data, to develop and validate a classification algorithm that could potentially improve the accuracy of ectopic pregnancy case identification.

METHODS

Using automated databases from two US managed-care plans, Group Health Cooperative (GH) and Kaiser Permanente Colorado (KPCO), we sampled women aged 15-44 with an ectopic pregnancy diagnosis or procedure code from 2001 to 2007 and verified their true case status through medical record review. We calculated positive predictive values (PPV) for code-selected cases compared with true cases at both sites. Using additional variables from the automated databases and classification and regression tree (CART) analysis, we developed a case-finding algorithm at GH (n = 280), which was validated at KPCO (n = 500).

RESULTS

Compared with true cases, the PPV of code-selected cases was 68 and 81% at GH and KPCO, respectively. The case-finding algorithm identified three predictors: ≥ 2 visits with an ectopic pregnancy code within 180 days; International Classification of Diseases, 9th Revision, Clinical Modification codes for tubal pregnancy; and methotrexate treatment. Relative to true cases, performance measures for the development and validation sets, respectively, were: 93 and 95% sensitivity; 81 and 81% specificity; 91 and 96% PPV; 84 and 79% negative predictive value. Misclassification proportions were 32% in the development set and 19% in the validation set when using standard codes; they were 11 and 8%, respectively, when using the algorithm.

CONCLUSIONS

The ectopic pregnancy algorithm improved case-finding accuracy over use of standard codes alone and generalized well to a second site. When using administrative data to select potential ectopic pregnancy cases, additional widely available automated health plan data offer the potential to improve case identification.

Electronic medical records (EMRs), epidemiology, and epistemology: reflections on EMRs and future pediatric clinical research

Electronic medical records (EMRs) are increasingly common in pediatric patient care. EMR data represent a relatively novel and rich resource for clinical research. The fact, however, that pediatric EMR data are collected for the purposes of clinical documentation and billing rather than research creates obstacles to their use in scientific investigation. Particular issues include accuracy, completeness, comparability between settings, ease of extraction, and context of recording. Although these problems can be addressed through standard strategies for dealing with partially accurate and incomplete data, a longer-term solution will involve work with pediatric clinicians to improve data quality. As research becomes one of the explicit purposes for which pediatricians collect EMR data, the pediatric clinician will play a central role in future pediatric clinical research.

Risks of congenital malformations and perinatal events among infants exposed to calcium channel and beta-blockers during pregnancy

PURPOSE

Calcium channel blockers and beta-blockers (BBs) are widely used during pregnancy, but data on their safety for the developing infant are scarce. We used population-based data from 5 HMOs to study risks for perinatal complications and congenital defects among infants exposed in-utero.

METHODS

We studied women older than 15 years delivering an infant between 1/1/96 and 12/31/00, who had been continuously enrolled with prescription drug coverage for ≥ 1 year prior to delivery. Information on prescription drug dispensings, inpatient, and outpatient diagnoses and procedures was obtained from automated databases at each HMO.

RESULTS

There were 584 full-term infants exposed during pregnancy to BBs and 804 full-term infants exposed to calcium-channel blockers, and over 75,000 unexposed mother-infant pairs with ≥ 30 days follow-up. Infants exposed to BBs in the third trimester of pregnancy had over threefold increased risk for hypoglycemia (RR 3.1; 95% CI 2.2, 4.2) and an approximately twofold increased risk for feeding problems (RR 1.8; 95% CI 1.3, 2.5). Infants exposed to calcium-channel blockers in the third trimester had an increased risk for seizures (RR 3.6 95% CI 1.3, 10.4). Chart review confirmed the majority of the exposed seizure and hypoglycemia cases. There were no increased risks for congenital anomalies among either group of infants, except for the category of upper alimentary tract anomalies; this increased risk was based on only two exposed cases.

CONCLUSIONS

Infants whose mothers receive BBs are at increased risk for neonatal hypoglycemia, while those whose mothers take calcium-channel blockers are at increased risk for neonatal seizures.

Combining electronic healthcare databases in Europe to allow for large-scale drug safety monitoring: the EU-ADR Project

PURPOSE

In this proof-of-concept paper we describe the framework, process, and preliminary results of combining data from European electronic healthcare record (EHR) databases for large-scale monitoring of drug safety.

METHODS

Aggregated demographic, clinical, and prescription data from eight databases in four countries (Denmark, Italy, Netherlands, the UK) were pooled using a distributed network approach by generation of common input data followed by local aggregation through custom-built software, Jerboa(©). Comparison of incidence rates of upper gastrointestinal bleeding (UGIB) and nonsteroidal anti-inflammatory drug (NSAID) utilization patterns were used to evaluate data harmonization and quality across databases. The known association of NSAIDs and UGIB was employed to demonstrate sensitivity of the system by comparing incidence rate ratios (IRRs) of UGIB during NSAID use to UGIB during all other person-time.

RESULTS

The study population for this analysis comprised 19,647,445 individuals corresponding to 59,929,690 person-years of follow-up. 39,967 incident cases of UGIB were identified during the study period. Crude incidence rates varied between 38.8 and 109.5/100,000 person-years, depending on country and type of database, while age-standardized rates ranged from 25.1 to 65.4/100,000 person-years. NSAID use patterns were similar for databases within the same country but heterogeneous among different countries. A statistically significant age- and gender-adjusted association between use of any NSAID and increased risk for UGIB was confirmed in all databases, IRR from 2.0 (95%CI:1.7-2.2) to 4.3 (95%CI: 4.1-4.5).

CONCLUSIONS

Combining data from EHR databases of different countries to identify drug-adverse event associations is feasible and can set the stage for changing and enlarging the scale for drug safety monitoring.

Privacy-maintaining propensity score-based pooling of multiple databases applied to a study of biologics

INTRODUCTION

A large study on the safety of biologics required pooling of data from multiple data sources, but while extensive confounder adjustment was necessary, private, individual-level covariate information could not be shared.

OBJECTIVES

To describe the methods of pooling data that investigators considered, and to detail the strengths and limitations of the chosen method: a propensity score (PS)-based approach that allowed for full multivariate adjustment without compromising patient privacy.

RESEARCH DESIGN

The project had a central data coordinating center responsible for collection and analysis of data. Private data could not be transmitted to the data coordinating center. Investigators assessed 4 methods for pooled analyses: full covariate sharing, cell-aggregated sharing, meta-analysis, and the PS-based method. We evaluated each method for protection of private information, analytic integrity and flexibility, and ability to meet the study's operational and statistical needs.

RESULTS

Analysis of 4 example datasets yielded substantially similar estimates if data were pooled with a PS versus individual covariates (0%-3% difference in point estimates). Several practical challenges arose. (1) PSs are best suited for dichotomous exposures but 6 or more exposure categories were desired; we chose a series of exposure contrasts with a common referent group. (2) Subgroup analyses had to be specified a priori. (3) Time-varying exposures and confounders required appropriate analytic handling including re-estimation of PSs. (4) Detection of heterogeneity among centers was necessary.

CONCLUSIONS

The PS-based pooling method offered strong protection of patient privacy and a reasonable balance between analytic integrity and flexibility of study execution. We would recommend its use in other studies that require pooling of databases, multivariate adjustment, and privacy protection.

Distributed health data networks: a practical and preferred approach to multi-institutional evaluations of comparative effectiveness, safety, and quality of care

BACKGROUND

Comparative effectiveness research, medical product safety evaluation, and quality measurement will require the ability to use electronic health data held by multiple organizations. There is no consensus about whether to create regional or national combined (eg, "all payer") databases for these purposes, or distributed data networks that leave most Protected Health Information and proprietary data in the possession of the original data holders.

OBJECTIVES

Demonstrate functions of a distributed research network that supports research needs and also address data holders concerns about participation. Key design functions included strong local control of data uses and a centralized web-based querying interface.

RESEARCH DESIGN

We implemented a pilot distributed research network and evaluated the design considerations, utility for research, and the acceptability to data holders of methods for menu-driven querying. We developed and tested a central, web-based interface with supporting network software. Specific functions assessed include query formation and distribution, query execution and review, and aggregation of results.

RESULTS

This pilot successfully evaluated temporal trends in medication use and diagnoses at 5 separate sites, demonstrating some of the possibilities of using a distributed research network. The pilot demonstrated the potential utility of the design, which addressed the major concerns of both users and data holders. No serious obstacles were identified that would prevent development of a fully functional, scalable network.

CONCLUSIONS

Distributed networks are capable of addressing nearly all anticipated uses of routinely collected electronic healthcare data. Distributed networks would obviate the need for centralized databases, thus avoiding numerous obstacles.

The Cardiovascular Research Network: a new paradigm for cardiovascular quality and outcomes research

BACKGROUND

A clear need exists for a more systematic understanding of the epidemiology, diagnosis, and management of cardiovascular diseases. More robust data are also needed on how well clinical trials are translated into contemporary community practice and the associated resource use, costs, and outcomes.

METHODS AND RESULTS

The National Heart, Lung, and Blood Institute recently established the Cardiovascular Research Network, which represents a new paradigm to evaluate the epidemiology, quality of care, and outcomes of cardiovascular disease and to conduct future clinical trials using a community-based model. The network includes 15 geographically distributed health plans with dedicated research centers, National Heart, Lung, and Blood Institute representatives, and an external collaboration and advisory committee. Cardiovascular research network sites bring complementary content and methodological expertise and a diverse population of approximately 11 million individuals treated through various health care delivery models. Each site's rich electronic databases (eg, sociodemographic characteristics, inpatient and outpatient diagnoses and procedures, pharmacy, laboratory, and cost data) are being mapped to create a standardized virtual data warehouse to facilitate rapid and efficient large-scale research studies. Initial projects focus on (1) hypertension recognition and management, (2) quality and outcomes of warfarin therapy, and (3) use, outcomes, and costs of implantable cardioverter defibrillators.

CONCLUSIONS

The Cardiovascular Research Network represents a new paradigm in the approach to cardiovascular quality of care and outcomes research among community-based populations. Its unique ability to characterize longitudinally large, diverse populations will yield novel insights into contemporary disease and risk factor surveillance, management, outcomes, and costs. The Cardiovascular Research Network aims to become the national research partner of choice for efforts to improve the prevention, diagnosis, treatment, and outcomes of cardiovascular diseases.

A distributed research network model for post-marketing safety studies: the Meningococcal Vaccine Study

PURPOSE

We describe a multi-center post-marketing safety study that uses distributed data methods to minimize the need for covered entities to share protected health information (PHI). Implementation has addressed several issues relevant to creation of a large scale post-marketing drug safety surveillance system envisioned by the FDA's Sentinel Initiative.

METHODS

This retrospective cohort study of Guillain-Barré syndrome (GBS) following meningococcal conjugate vaccination incorporates the data and analytic expertise of five research organizations closely affiliated with US health insurers. The study uses administrative claims data, plus review of full text medical records to adjudicate the status of individuals with a diagnosis code for GBS (ICD9 357.0). A distributed network approach is used to create the analysis files and to perform most aspects of the analysis, allowing nearly all of the data to remain behind institutional firewalls. Pooled analysis files transferred to a central site will contain one record per person for approximately 0.2% of the study population, and contain PHI limited to the month and year of GBS onset for cases or the index date for matched controls.

RESULTS

The first planned data extraction identified over 9 million eligible adolescents in the target age range of 11-21 years. They contributed an average of 14 months of eligible time on study over 27 months of calendar time. MCV4 vaccination coverage levels exceeded 20% among 17-18-year olds and 16% among 11-13 and 14-16-year-old age groups by the second quarter of 2007.

CONCLUSION

This study demonstrates the feasibility of using a distributed data network approach to perform large scale post-marketing safety analyses and is scalable to include additional organizations and data sources. We believe these results can inform the development of a large national surveillance system.

Partnerships in translation: advancing research and clinical care. The 14th Annual HMO Research Network Conference, April 13-16, 2008, Minneapolis, Minnesota

The Health Maintenance Organization Research Network held its annual meeting in Minneapolis in April of 2008, with more than 300 investigators, research staff, clinical leaders, and academic partners gathering in conjunction with the conference theme 'Partnerships in Translation: Advancing Research and Clinical Care.' This article provides some background on the network, its research activities, and the annual conference. Also featured is an article by Coleman and colleagues summarizing the conference's first plenary session, where operational leaders of health care organizations discussed the optimization of health care through research. This issue of Clinical Medicine & Research also includes a selection of scientific abstracts presented at the meeting on a wide range of clinical and population health topics.

Hormone therapy initiation after the Women's Health Initiative

OBJECTIVES

To describe hormone therapy (HT) initiation after the 2002 publication of the Women's Health Initiative.

DESIGN

Observational cohort (1999-2003) of women ages 40 to 79 years, five health plans, used HT in July 2002 and subsequently discontinued or never used before August 2002.

RESULTS

Of discontinuers, 15.8% (3,203 of 20,205) reinitiated HT. Reinitiation was higher among estrogen users (23.8%) versus estrogen with progestin users (11.3%), and lower among those with diabetes (relative risk [RR]=0.68, 95% CI: 0.61-0.76), cardiovascular disease (RR=0.87, 95% CI: 0.83-0.92), and hyperlipidemia (RR=0.83, 95% CI: 0.79-0.88). Only 2.3% (2,072 of 90,261) of never users initiated (August 2002 to December 2003). First-time initiation was associated with cardiovascular disease (RR=1.17, 95% CI: 1.10-1.25) and hyperlipidemia (RR=1.24, 95% CI: 1.17-1.33) and was less common among those with diabetes (RR=0.70, 95% CI: 0.63-0.79).

CONCLUSIONS

After the Women's Health Initiative, a minority of women reinitiated or became first-time initiators of HT. Women with cardiovascular disease, diabetes, and hyperlipidemia were less likely to reinitiate; women with cardiovascular disease and hyperlipidemia were more likely to be first-time initiators.

Asthma drug use and the development of Churg-Strauss syndrome (CSS)

PURPOSE

Case reports suggest that leukotriene modifier use may be associated with the onset of Churg-Strauss syndrome (CSS). Using pooled data from two nested case-control studies, we examined the association between asthma drug use and the development of CSS.

METHODS

The study was performed in three US managed care organizations and a US national health plan with chart access and complete electronic pharmacy data, with a covered population of 13.9 million. There were 47 cases of possible or definite CSS and 4700 asthma drug user controls identified between January 1, 1995 and December 31, 2002. We examined exposure to asthma drugs in cases and controls, including leukotriene modifiers (6 cases and 202 controls), in the two to 6 months prior to the onset of adjudicated CSS.

RESULTS

While the crude association between use of leukotriene modifiers and CSS was strong (odds ratio (OR) 4.00, 95% confidence interval (CI): 1.49-10.60), in a multivariable analysis controlling for use of oral corticosteroids, inhaled corticosteroids, and number of categories of asthma drugs dispensed, there was no significant association (OR 1.32, 95% CI: 0.44-3.96). Use of inhaled and oral corticosteroids, evaluated as markers of asthma severity, were associated with CSS (OR 3.07, 95% CI: 1.34-7.03 and OR 5.36, 95% CI: 2.51-11.45, respectively).

CONCLUSIONS

No association was found between CSS and leukotriene modifiers after controlling for asthma drug use However, it is not possible to rule out modest associations with asthma treatments given CSS is so rare and so highly correlated with asthma severity, suggesting further investigation is warranted.

The Emergence of Biobanks: Practical Design Considerations for Large Population-Based Studies of Gene-Environment Interactions

The completion of the human genome project has spurred new thinking about launching large-scale cohort studies; as proposed, these studies will differ from past large-scale cohort studies and will focus primarily on how genetic variation interacts with environmental exposures to affect the risk for common human diseases. There is no single 'best design' for large-scale studies of gene-environment interactions. Some studies are best performed in cohort studies where unbiased information can be collected on individuals years before disease onset. Other studies may be most efficiently done with a case-control design using currently available automated data. Population-based biobanks with nested case-control or case-cohort studies offer distinct advantages to some of the resource-intensive large-scale cohort studies under consideration, and may be more acceptable to many of the countries around the world currently considering such projects.

Validity of gout diagnoses in administrative data

OBJECTIVE

To determine the utility of using administrative data for epidemiologic studies of gout by examining the validity of gout diagnoses in claims data.

METHODS

From a population of approximately 800,000 members from 4 managed care plans, we identified patients who had at least 2 ambulatory claims for a diagnosis of gout between January 1, 1999 and December 31, 2003. From this group, a random sample of 200 patients was chosen for medical record review. Trained medical record reviewers abstracted gout-related clinical, laboratory, and radiologic data from the medical records. Two rheumatologists independently evaluated the abstracted information and assessed whether the gout diagnosis was probable/definite or unlikely/insufficient information. Discordant physician ratings were adjudicated by consensus. Based on record reviews, patients were also classified according to the American College of Rheumatology (ACR), Rome, and New York gout criteria and these results were compared with the physician global assessments.

RESULTS

There were 121 patients rated as having probable/definite gout by physician consensus, leading to a positive predictive value of >or=2 coded diagnoses of gout of 61% (95% confidence interval 53-67). There was low concordance between physician assessments and established gout criteria including ACR, Rome, and New York criteria (kappa = 0.17, 0.16, and 0.20, respectively).

CONCLUSION

Use of administrative data alone in epidemiologic and health services research on gout may lead to misclassification. Medical record reviews for validation of claims data may provide an inadequate gold standard to confirm gout diagnoses.

Curbing The Cardiovascular Disease Epidemic: Aligning Industry, Government, Payers, And Academics

Despite decades of progress in the diagnosis, treatment, and prevention of cardiovascular disease, its prevalence continues to grow in both developed and developing countries. We have constructed a model, the "cycle of quality," which connects the innovation of initial scientific discovery with validated methods of translating research into effective delivery. This model can serve as a basis for evaluating proposed efforts to improve interactions among private and public aspects of health care to accelerate development and appropriate adoption of new treatments, and to achieve greater penetration of effective behavioral therapies and established technologies, resulting in major improvements in cardiovascular health.

Sex differences in gout epidemiology: evaluation and treatment

BACKGROUND

Little is known about the characteristics, evaluation and treatment of women with gout.

OBJECTIVE

To examine the epidemiological differences and differences in treatment between men and women in a large patient population.

METHODS

The data from approximately 1.4 million people who were members of seven managed care plans in the USA for at least 1 year between 1 January 1999 and 31 December 2003 were examined. Adult members who had pharmacy benefits and at least two ambulatory claims specifying a diagnosis of gout were identified. In addition, men and women who were new users of urate-lowering drugs (ULDs) were identified to assess adherence with recommended surveillance of serum urate levels within 6 months of initiating urate-lowering treatment.

RESULTS

A total of 6133 people (4975 men and 1158 women) with two or more International Classification of Disease-9 codes for gout were identified. As compared with men with gout, women were older (mean age 70 (SD 13) v 58 (SD 14), p<0.001) and had comorbidities and received diuretics more often (77% v 40%; p<0.001). Only 37% of new users of urate-lowering treatment had appropriate surveillance of serum urate levels post-initiation of urate-lowering treatment. After controlling for age, comorbidities, gout treatments, number of ULD dispensings and health plan, women were more likely (odds ratio 1.36, 95% confidence interval 1.11 to 1.67) to receive the recommended serum urate level testing.

CONCLUSIONS

Women with gout were older, had greater comorbidities and more often used diuretics and received appropriate surveillance of serum urate levels, suggesting that the factors leading to gout as well as monitoring of treatment are very different in women and men.

Reliability of Group Health Cooperative automated pharmacy data by drug benefit status

PURPOSE

Evaluate the reliability of health plan pharmacy records in determining medication use among seniors with and without a drug benefit.

METHODS

Subjects included 3610 seniors, enrolled in Group Health Cooperative's Medicare (GHC) + Choice program during 1998-1999, receiving care in an integrated group practice (IGP), and diagnosed with one or more of four chronic conditions (hypertension, diabetes, congestive heart failure, and coronary artery disease). We compared pharmacy records to self-reported medication use for antidepressant, antihypertensive, acid suppressant, cardiac, diabetic, hormone, and lipid lowering drugs.

RESULTS

Agreement between pharmacy records and self-report was substantial to almost perfect (prevalence-adjusted and bias-adjusted kappa (PABAK) range: 0.69 for antihypertensives to 0.95 for cardiac agents) among seniors with a drug benefit. Agreement was slightly less for seniors without a drug benefit (PABAK range: 0.51 for antihypertensives to 0.92 for cardiac agents) and differences varied by drug class. Among seniors without a benefit, the prevalence of medication use was lower when based on pharmacy records than when based on self-report for all medication classes of interest.

CONCLUSIONS

While GHC may not be representative of all health plans, our study indicates that health plan pharmacy records are a reliable source of data for seniors receiving care within an IGP. However, the reliability of pharmacy records appears better among seniors with a drug benefit. Researchers should consider factors such as drug benefit status when conducting studies using pharmacy data. More studies are needed in different populations and delivery systems, as well as over varied types of drug benefits.

Practice-level effects of interventions to improve asthma care in primary care settings: the Pediatric Asthma Care Patient Outcomes Research Team

OBJECTIVE

To assess the practice-level effects of (1) a physician peer leader intervention and (2) peer leaders in combination with the introduction of asthma education nurses to facilitate care improvement. And, to compare findings with previously reported patient-level outcomes of trial enrollees.

STUDY SETTING

Data were included on children 5-17 years old with asthma in 40 primary care practices, affiliated with managed health care plans enrolled in the Pediatric Asthma Care Patient Outcomes Research Team (PORT) randomized trial.

STUDY DESIGN

Primary care practices were randomly assigned to one of two care improvement arms or to usual care. Automated claims data were analyzed for 12-month periods using a repeated cross-sectional design. The primary outcome was evidence of at least one controller medication dispensed among patients with persistent asthma. Secondary outcomes included controller dispensing among all identified asthmatics, evidence of chronic controller use, and the dispensing of oral steroids. Health service utilization outcomes included numbers of ambulatory visits and hospital-based events.

PRINCIPAL FINDINGS

The proportion of children with persistent asthma prescribed controllers increased in all study arms. No effect of the interventions on the proportion receiving controllers was detected (peer leader intervention effect 0.01, 95 percent confidence interval [CI]: -0.07, 0.08; planned care intervention effect -0.03, 95 percent CI: -0.09, 0.02). A statistical trend was seen toward an increased number of oral corticosteroid bursts dispensed in intervention practices. Significant adjusted increases in ambulatory visits of 0.08-0.10 visits per child per year were seen in the first intervention year, but only a statistical trend in these outcomes persisted into the second year of follow-up. No differences in hospital-based events were detected.

CONCLUSIONS

This analysis showed a slight increase in ambulatory asthma visits as a result of asthma care improvement interventions, using automated data. The absence of detectable impact on medication use at the practice level differs from the positive intervention effect observed in patient self-reported data from trial enrollees. Analysis of automated data on nonenrollees adds information about practice-level impact of care improvement strategies. Benefits of practice-level interventions may accrue disproportionately to the subgroup of trial enrollees. The effect of such interventions may be less apparent at the level of practices or health plans.

Measuring and improving performance in multicenter research consortia

BACKGROUND

Some evidence suggests that the quality of the organization and management of research consortia influences productivity and staff satisfaction. Collaborators in a research consortium generally focus on developing and implementing studies and thus rarely assess the process of collaboration. We present an approach to evaluating and improving a research consortium, using the HMO Cancer Research Network (CRN) as an example.

METHODS

Five domains are evaluated: extent of collaboration and quality of communication; performance of projects and infrastructure; data quality; scientific productivity; and impact on member organizations. The primary assessment tool is a survey of CRN scientists and project staff, undertaken annually.

RESULTS

Each year, the evaluation has identified critical aspects of this collaboration that could be improved. Several tangible changes have been implemented to improve productivity of the consortium. The most important result of the CRN Evaluation is the ability to have open dialogue about ways to improve its overall performance.

CONCLUSION

Optimizing the process of collaboration will contribute to achievement of the scientific goals. The experience of the CRN provides a useful framework and process for evaluating the structure of consortium-based research.

A review of uses of health care utilization databases for epidemiologic research on therapeutics

OBJECTIVE

Large health care utilization databases are frequently used in variety of settings to study the use and outcomes of therapeutics. Their size allows the study of infrequent events, their representativeness of routine clinical care makes it possible to study real-world effectiveness and utilization patterns, and their availability at relatively low cost without long delays makes them accessible to many researchers. However, concerns about database studies include data validity, lack of detailed clinical information, and a limited ability to control confounding.

STUDY DESIGN AND SETTING

We consider the strengths, limitations, and appropriate applications of health care utilization databases in epidemiology and health services research, with particular reference to the study of medications.

CONCLUSION

Progress has been made on many methodologic issues related to the use of health care utilization databases in recent years, but important areas persist and merit scrutiny.

The impact of comorbidities on hormone use. After the 2000 release of the Women's Health Initiative

OBJECTIVE

Determine the impact of fracture, coronary disease, and diabetes on changes in rates of discontinuation and initiation of estrogen therapy with (EPT) and without (ET) progestin, before (September 1, 1999 to June 30, 2002, baseline) versus 5 months after (follow-up) release of the Women's Health Initiative EPT trial results (WHI).

DESIGN, SETTING, AND PARTICIPANTS

Observational cohort; 169,586 women 40 to 80 years old from 5 U.S. HMOs.

METHODS

We used pharmacy data to identify ET and EPT users. A woman was a user any month she filled > or =1 estrogen prescription and in subsequent months based upon the number of pills/patches dispensed. We used inpatient and outpatient claims to identify fracture January 1, 1999 to June 30, 2002 and pharmacy data to identify disease-based groups of medications for diabetes and cardiovascular disease.

MEASURES

EPT/ET prevalence, initiation, and discontinuation rates.

RESULTS

Baseline to follow-up EPT and ET prevalence declined 45% and 22%, respectively, with no difference by comorbidity. Follow-up EPT initiation was half the baseline rate irrespective of comorbidity. Compared to baseline, follow-up EPT discontinuation rates increased among women with diabetes (relative risk [RR], 6.9; 95% confidence interval [CI], 5.6 to 8.4), cardiovascular disease (RR, 5.5; 95% CI, 4.9 to 6.2), fracture (RR, 3.8; 95% CI, 2.4 to 5.7), and no comorbidity (RR, 4.4; 95% CI, 3.9 to 4.9). The RRs for follow-up versus baseline EPT discontinuation were higher among women with diabetes (P<.01) and cardiovascular disease (P<.01) versus women without these comorbidities. ET discontinuation rates among these same groups were elevated 2- to 2.8-fold.

CONCLUSIONS

Diabetes and cardiovascular disease were associated with higher EPT discontinuation rates post-WHI compared to women without comorbidity; comorbidity had little impact on changes in prevalence or initiation of ET/EPT after release of the WHI.