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.

Diffusion of aromatase inhibitors for breast cancer therapy between 1996 and 2003 in the Cancer Research Network.

PURPOSE: Clinical trials demonstrated adjuvant aromatase inhibitor treatment is superior for decreasing breast cancer recurrence risk over adjuvant tamoxifen treatment as early as 2001. Yet clinical use for adjuvant treatment was not recommended by the American Society of Clinical Oncology until 2004. Aromatase inhibitor uptake after the first public presentation of randomized trial results but before the release of national guidelines is unclear. We evaluated diffusion of aromatase inhibitor dispensings for breast cancer treatment in integrated healthcare delivery systems across the United States. METHODS: We collected automated data for 13,245 women enrolled at seven integrated healthcare delivery systems in the Cancer Research Network. All women were aged >55 and diagnosed with estrogen receptor positive, invasive breast cancer between 1996 and 2003. We used electronic pharmacy data to identify aromatase inhibitor and tamoxifen dispensings through 2004. We evaluated the proportions of women who received hormone dispensings in two ways: (1) at any point after diagnosis to capture all use, and (2) in the two-year period following diagnosis to approximate adjuvant use. RESULTS: Over time, adjuvant aromatase inhibitor use increased whereas tamoxifen use decreased. Aromatase inhibitor dispensings within 2 years of diagnosis increased from 4.1% among women diagnosed in 2000 to 13% in 2001, 24% in 2002, and 40% in 2003. Tamoxifen use declined starting in 2001 at every system. CONCLUSION: Aromatase inhibitor use rose dramatically after 2001 while tamoxifen use decreased. It appears results from early clinical trials changed practice in these integrated healthcare systems before formal changes in national guidelines.

Use of anti-inflammatory drugs and lower esophageal sphincter-relaxing drugs and risk of esophageal and gastric cancers.

BACKGROUND & AIMS: The incidence of esophageal and gastric cardia adenocarcinoma has increased in Western countries in recent decades for largely unknown reasons. We investigated whether use of LES-relaxing drugs was related to an increased risk of esophageal and gastric cardia adenocarcinoma, and whether use of NSAIDs was related to a reduced risk of esophageal and gastric cancers. METHODS: We examined these associations by using administrative databases in a case-control study in 2 integrated health care delivery systems. Cases were incident esophageal adenocarcinomas (n = 163) and squamous cell carcinomas (n = 114) and gastric cardia (n = 176) and non-cardia adenocarcinomas (n = 320), diagnosed between 1980-2002 in one health system and between 1993-2002 in the other. Matched controls (n = 3996) were selected. Complete prescription information was available for the study period. RESULTS: Prescription of corticosteroids was associated with a decreased risk of esophageal adenocarcinoma (odds ratio [OR], 0.6; 95% confidence interval [CI], 0.4-0.9), esophageal squamous cell carcinoma (OR, 0.4; 95% CI, 0.2-0.6), and gastric non-cardia carcinoma (OR, 0.4, 95% CI, 0.3-0.6). Ever use of pharmacy-purchased aspirin was associated with 30%-60% decreased risks of the studied cancers. As a group, LES-relaxing drugs showed little evidence of association with increased risk of any esophageal or gastric cancer. CONCLUSIONS: Corticosteroid and aspirin use were associated with significantly decreased risks of esophageal and gastric cancer. LES-relaxing drugs as a group did not affect these risks, although we had limited power to assess individual drugs. The possibility that corticosteroids and aspirin might reduce esophageal cancer risk warrants further consideration.

Racial differences in tumor stage and survival for colorectal cancer in an insured population.

BACKGROUND: Despite declining death rates from colorectal cancer (CRC), racial disparities have continued to increase. In this study, the authors examined disparities in a racially diverse group of insured patients. METHODS: This study was conducted among patients who were diagnosed with CRC from 1993 to 1998, when they were enrolled in integrated healthcare systems. Patients were identified from tumor registries and were linked to information in administrative databases. The sample was restricted to non-Hispanic whites (n = 10,585), non-Hispanic blacks (n = 1479), Hispanics (n = 985), and Asians/Pacific Islanders (n = 909). Differences in tumor stage and survival were analyzed by using polytomous and Cox regression models, respectively. RESULTS: In multivariable regression analyses, blacks were more likely than whites to have distant or unstaged tumors. In Cox models that were adjusted for nonmutable factors, blacks had a higher risk of death from CRC (hazard ratio [HR], 1.17; 95% confidence interval [95% CI], 1.06-1.30). Hispanics had a risk of death similar to whites (HR, 1.04; 95% CI, 0.92-1.18), whereas Asians/Pacific Islanders had a lower risk of death from CRC (HR, 0.89; 95% CI, 0.78-1.02). Adjustment for tumor stage decreased the HR to 1.11 for blacks, and the addition of receipt of surgical therapy to the model decreased the HR further to 1.06. The HR among Hispanics and Asians/Pacific Islanders was stable to adjustment for tumor stage and surgical therapy. CONCLUSIONS: The relation between race and survival from CRC was complex and appeared to be related to differences in tumor stage and therapy received, even in insured populations. Targeted interventions to improve the use of effective screening and treatment among vulnerable populations may be needed to eliminate disparities in CRC.

Building a research consortium of large health systems: the Cancer Research Network.

Critical questions about cancer prevention, care, and outcomes increasingly require research involving large patient populations and their care delivery organizations. The Cancer Research Network (CRN) includes 11 integrated health systems funded by the National Cancer Institute (NCI) to conduct collaborative cancer research. This article describes the challenges of constructing a productive consortium of large health systems, and explores the CRN's responses. The CRN was initially funded through an NCI cooperative agreement in 1999 and has since received a second 4-year grant. Leadership and policy development are provided through a steering committee, subcommittees, and an external advisory committee. The CRN includes integral and affiliated research projects supported by a Scientific and Data Resources Core. Three characteristics of the CRN intensified the general challenges of consortium research: 1) its members are large health systems with legitimate concerns about confidentiality of data about enrollees, providers, and the organization; 2) CRN research projects often generate highly sensitive data about quality of care; and therefore 3) each participating organization wants a strong voice in CRN direction. CRN experience to date confirms that a consortium of health systems with internal research capacity can address a range of important cancer research questions that would be difficult to study in other venues. The advantages and challenges of consortium research are explored, with suggestions for the development, execution, and management of multisystem population laboratories.

Building a virtual cancer research organization.

BACKGROUND: The Cancer Research Network (CRN) comprises the National Cancer Institute and 11 nonprofit research centers affiliated with integrated health care delivery systems. The CRN, a public/private partnership, fosters multisite collaborative research on cancer prevention, screening, treatment, survival, and palliation in diverse populations. METHODS: The CRN's success hinges on producing innovative cancer research that likely would not have been developed by scientists working individually, and then translating those findings into clinical practice within multiple population laboratories. The CRN is a collaborative virtual research organization characterized by user-defined sharing among scientists and health care providers of data files as well as direct access to researchers, computers, software, data, research participants, and other resources. The CRN's research management Web site fosters a high-functioning virtual scientific community by publishing standardized data definitions, file specifications, and computer programs to support merging and analyzing data from multiple health care systems. RESULTS: Seven major types of standardized data files developed to date include demographics, health plan eligibility, tumor registry, inpatient and ambulatory utilization, medication dispensing, laboratory tests, and imaging procedures; more will follow. Data standardization avoids rework, increases multisite data integrity, increases data security, generates shorter times from initial proposal concept to submission, and stimulates more frequent collaborations among scientists across multiple institutions. CONCLUSIONS: The CRN research management Web site and associated standardized data files and procedures represent a quasi-public resource, and the CRN stands ready to collaborate with researchers from outside institutions in developing and conducting innovative public domain research.

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.

Disparities and survival among breast cancer patients.

BACKGROUND: Although rates of survival for women with breast cancer have improved, the survival disparity between African American and white women in the United States has increased. PURPOSE: To determine whether this survival disparity persists in an insured population with access to medical care. METHODS: In this retrospective cohort study, we extracted data from the tumor registries of six nonprofit, integrated health care delivery systems affiliated with the Cancer Research Network and assessed the survival of African American (n = 2276) and white (n = 18 879) female enrollees who were diagnosed with invasive breast cancer from January 1, 1993, through December 31, 1998. Cox proportional hazards regression was used to estimate the death rate among African American women relative to that of white women after adjustment for potential explanatory factors including stage at diagnosis, tumor characteristics, and treatment. RESULTS: Five-year survival was lower for African American women (73.8%) than for white women (81.6%). African American women were less likely to have tumor characteristics with good prognosis. Controlling for age at diagnosis, stage, grade, tumor size, and estrogen and progesterone receptor status, the adjusted hazard rate ratio for African American women was 1.34 (95% confidence interval = 1.22 to 1.46). Similar risks were found among women ages 20-49 and 50 and older. Controlling for treatment slightly lowered the hazard rate ratio to 1.31 (95% confidence interval = 1.20 to 1.43). CONCLUSIONS: Among women with invasive breast cancer, being insured and having access to medical care does not eliminate the survival disparity for African American women.

Efficacy of breast cancer screening in the community according to risk level.

BACKGROUND: The efficacy of breast cancer screening in the community may differ from that suggested by the results of randomized trials, and no data have been available on efficacy among women who have different levels of breast cancer risk. METHODS: We conducted a matched case-control study among women enrolled in six health plans in Washington, Oregon, California, Massachusetts, and Minnesota. We examined the efficacy of screening by mammography and/or clinical breast examination among women in two age cohorts (40-49 years and 50-65 years) and in two breast cancer risk levels (average and increased risk). Women who died from breast cancer from January 1, 1983, through December 31, 1998, (N = 1351; case subjects) were matched to control subjects (N = 2501) on age and risk level. Increased risk was defined as a family history of breast cancer or a breast biopsy noted in the medical records before the index date (defined as date of first suspicion of breast abnormalities in case subjects, with the same date used for matched control subjects). Data on screening, risk status, and other variables were abstracted from medical records. Conditional logistic regression was used to examine the association between breast cancer mortality and receipt of screening. All statistical tests were two-sided. RESULTS: There were small, non-statistically significant associations between breast cancer mortality and receipt of screening during the 3 years prior to the index date for both the younger women [odds ratio (OR) = 0.92; 95% confidence interval (CI) = 0.76 to 1.13] and the older women (OR = 0.87; 95% CI = 0.68 to 1.12). The association among women at increased risk (OR = 0.74; 95% CI = 0.50 to 1.03) was stronger than that among women at average risk (OR = 0.96; 95% CI = 0.80 to 1.14), but the difference was not statistically significant (P = .17). CONCLUSIONS: In this community-based study, screening history was not associated with breast cancer mortality. However, potential limitations of this study argue for a cautious interpretation of these findings.

Comparing breast cancer case identification using HMO computerized diagnostic data and SEER data.

OBJECTIVE: To determine the sensitivity and positive predictive value (PPV) of computerized diagnostic data from health maintenance organizations (HMOs) in identifying incident breast cancer cases. STUDY DESIGN: An HMO without a cancer registry developed an algorithm identifying incident breast cancer cases using computerized diagnostic codes. Two other HMO sites with Surveillance, Epidemiology, and End Results (SEER) registries duplicated this case-identification approach. Using the SEER registries as the criterion standard, we determined the sensitivity and PPV of the computerized data. METHODS: Data were collected from HMO computerized data-bases between January 1, 1996, and December 31, 1999. Surveillance, Epidemiology, and End Results data were also used. RESULTS: The overall sensitivity of the HMO databases was between 0.92 (95% confidence interval [CI], 0.91-0.96) and 0.99 (95% CI, 0.98-0.99). Sensitivity was high (range, 0.94-0.98), for the first 3 (of 4) years, dropping slightly (range, 0.81-0.94) in the last year. The overall PPV ranged from 0.34 (95% CI, 0.32-0.35) to 0.44 (95% CI, 0.42-0.46). Positive predictive value rose sharply (range, 0.18-0.20) after the first year to 0.83 and 0.92 in the last year because prevalent cases were excluded. Review of a random sample of 50 cases identified in the computerized data-bases but not by SEER data indicated that, while SEER usually identified the cases, the registry did not associate every case with the health plan. CONCLUSIONS: Health maintenance organization computerized databases were highly sensitive for identifying incident breast cancer cases, but PPV was low in the initial year because the systems did not differentiate between prevalent and incident cases. Health maintenance organizations depending solely on SEER data for cancer case identification will miss a small percentage of cases.

Retention of enrollees following a cancer diagnosis within health maintenance organizations in the Cancer Research Network.

Population laboratories with complete clinical information on episodes of care are needed to support research on the quality of care delivered to cancer patients. Data resources within the Cancer Research Network (CRN) may overcome many of the limitations of existing cancer databases, but their potential clinical value depends on the stability of the enrolled population. To assess this issue, we studied the retention rates among survivors of the 132 580 patients diagnosed with cancer from January 1, 1993, through December 31, 1998, who were enrolled at five health maintenance organization sites participating in the CRN. Enrollees were followed from cancer diagnosis through death, disenrollment, or the end of follow-up (i.e., December 31, 1999). The retention rate among survivors for all cancers combined at 1 and 5 years after cancer diagnosis was 96.0% (95% confidence interval [CI] = 95.9% to 96.1%) and 83.9% (95% CI = 83.4% to 84.3%), respectively. The proportion of enrollees diagnosed with cancer who remained enrolled and available for evaluation suggests that the CRN is well-suited for studies of the quality of care for cancer patients, survivorship, and long-term outcomes.

A computerized system to facilitate medical record abstraction in cancer research (United States).

OBJECTIVE: To implement a computerized system to gather and transmit medical record information from six sites to a centralized database for two cancer prevention studies. METHODS: Microsoft Access 97 was selected as the application for the system. Sites purchased Access and hardware meeting technical specifications required for the system. A developer worked with the lead investigator and medical record abstractors to develop a 'back-end' database to hold the desired data while maintaining a user-friendly 'front-end' interface. Abstractors trained on a paper version of the abstraction form were then trained to use the system. Meeting minutes and technical notes were used in summarizing the approach and process. Observations were collected through discussions. RESULTS: We overcame multiple obstacles to develop computerized systems supporting medical record data collection at multiple sites. Although system development slowed implementation of the study, the system produced data for cleaning and analysis immediately. Overall the approach decreased the time from study implementation to manuscript submission. Development time for a second system was substantially reduced. CONCLUSIONS: Computerized systems for medical record abstraction at multiple sites convey substantial benefit. We present a schematic approach to facilitate development of similar systems in the future.