Currently, BHN’s HCAP data warehouse consists of clinical and administrative data from the eight Logician-using health centers, the Epicare-using health center, and BMC. Data is compiled from individual institutional servers into a Microsoft SQL Server using loading scripts, programs which automatically and regularly update data, developed as part of the HCAP grant. These scripts allow the health centers and hospital to refresh data on a regular basis at no incremental cost. The data warehouse supplies valuable patient tracking and utilization data to executive directors via monthly reports. With the exception of data specific to diabetic patients, the warehouse currently only contains administrative data extracted from practice management databases, however, the HCAP staff is working to build an Oracle relational database that will combine EMR clinical data with cost and demographic information. In the paragraphs below we describe key benefits and challenges associated with the HCAP data warehouse.
Design, implementation and maintenance challenges.Although the data warehouse represents an important resource as a central repository of demographic, utilization and cost data for uninsured patients receiving care at BHN health centers, HCAP respondents described significant challenges associated with its development and maintenance. First, an intense programming effort was required to design the warehouse as an index of unduplicated patient records and substantial effort is required to maintain the integrity of these records as data are added. In addition, HCAP staff faced considerable challenges in designing appropriate scripts to extract data across a myriad of servers and databases. As with other data integration activities described in this report, the development of custom data extracts was delayed because of reluctance on the part of practice management vendors to share data specifications and the high cost of using outside vendors to program and implement these extracts.
Finally, assessing data reliability and validity in the data warehouse has been an ongoing challenge for HCAP staff. The warehouse contains hundreds of individual variables relevant to each patient, most based on claims-type administrative data. HCAP staff conducts detailed validity and reliability checks on each variable prior to including them in reports and analyses. In many cases, variables are simply not used due to ambiguous data quality.
Utilization and cost reporting. Currently, HCAP staff produce a series of standard monthly reports of basic demographic, utilization and cost data by site. These reports provide each participating health center with a breakdown of demographics for uninsured patients; analysis of most frequent (top 10) diagnoses; utilization patterns for high cost patients; ED utilization for their patients and analysis of admissions for ambulatory care sensitive conditions. Over time, HCAP staff report they will work closely with health center leadership to refine existing reports and produce new reports to meet health center management requirements.
Uses of warehouse data. The data warehouse supports a range of stakeholder information needs. Health center administrators cited wide use of existing data warehouse reports to set policy and reallocate resources. Medical directors anticipate that these data will allow them to work with public health officials to improve provision of diagnostic and preventative services to vulnerable populations. All respondents indicated that these data are particularly useful given the lack of other sources on health care utilization and costs specific to the uninsured. Recently, HCAP has initiated collaboration with the state’s Department of Healthcare Financing and Policy so that these data can inform statewide programs and policies.
Finally, HCAP leadership are very interested in the potential for the data warehouse to contribute to research efforts. For example, HCAP submitted an R01 application to the Agency for Healthcare Research and Quality (AHRQ) in order to research the effect of discharge planning on re-admission rates. Additionally, they plan to use warehouse data to measure a baseline of mammography referrals across the network.
Diabetes specific reports. One area where we found evidence that the data warehouse efforts have successfully leveraged EMR data to improve patient care is in the area of diabetes. As described above, several health centers currently work with EMR diabetes templates as to support disease management. Data on the diabetic population is currently the only clinical data included as part of the data warehouse. Detailed reports allow tracking of key outcome measures such as HbA1c and blood pressure on a health center and provider level. Participating health centers reported observing tangible benefits as providers worked to improve on trends demonstrated in these reports.