Novel data sources for quality measurement

On Thursday, I'll give a presentation to the National Committee on Vital and Health Statistics (NCVHS) about measuring quality using "traditional" and emerging, novel sources of healthcare data.

My definition of traditional data sources that are currently used to measure quality includes administrative claims data aggregated from hospital-based claims databses (for example, BIDMC has an Oracle respository called Casemix), payer-based databases (all have a claims warehouse to support disease management), physician organizations (Beth Israel Deaconess Physicians Organization has worked with Heathcare Data Services to create all payer business intelligence tools ) and health data consortia (such as the Massachusetts Health Data Consortium offers de-identified aggregated claims to enable institutional comparisons)"

New sources of data for quality analysis go beyond administrative data and include EHR, PHR, and Healthcare Information Exchange resources. Here are a few examples:

*At BIDMC all our laboratory, radiology, pharmacy and care process data is available in business intelligence datamarts. We use these internally for scorecards, benchmarking and workflow improvement projects.

*Massachusetts has a long history of payer/provider collaboration, such as NEHEN. Recently, the Eastern Massachusetts Healthcare Initiative (EMHI) has developed a new set of clinical data exchange use cases to support regional payer/provider collaboration. One of those use cases is the automated exchange of quality data via a secure publish/subscribe web service that eliminates the need for providers to create bulk data extracts for payer quality measures.

*The Massachusetts eHealth Collaborative has created a quality data warehouse for its 600 participating clinicians. This warehouse is so good that the BIDMC Physicians Organization (BIDPO) has elected to use it for aggregating the quality measure data on its physicians.

*Surescripts/Rx Hub provides national medication list data that is helpful for clinical care and quality measurement

*Commercial labs such as Quest and LabCorps are implementing HITSP standards for lab transactions which include the data elements needed for biosurveillance, public health reporting and quality analysis.

*The Social Security Administration's Megahit pilot demonstrated automated submission of electronic medical records between hospitals and the SSA with patient consent to improve turn around time for disability claims adjudication.

*The Centers for Disease Control has implemented Biosense, an automated surveillance system for detecting variations in disease frequency using de-identified emergency department and hospital data.

*The Massachusetts Medical Society is working with the Massachusetts eHealth Collaborative (MAeHC) to pilot quality scorecards for its members using the MAeHC quality warehouse.

*Departments of Public Health in Massachusetts receive automated data feeds from local hospitals to enable early detection of outbreaks

*The AEGIS system developed by Children's hospital uses automated data feeds from Massachusetts hospitals to create real time influenza prevalence maps

*A new generation of consumer healthcare devices from the Continua Alliance enables remote monitoring of patients in the home, transmitting data to EHRs and PHRs.

*Personal Health Records includes those tethered to an EHR, those sponsored by employers, hosted by health plans and vendor-based systems such as Microsoft HealthVault and Google Health enable patients to aggregate, enter, and manage their own data. PHRs may be an appropriate way to measure quality by asking the patients to subscribe/contribute their data to quality measurement organizations. The trusted third party model in Google enables patients to share data with their consent. Some patients may feel altruistic enough to contribute their data for quality measurement.

*Google Trends shows search term trends over time. This can be used to quantify searches on symptoms such as flu-related illnesses, providing early detection of changes in the frequency of users searching for fever/cough/flu etc. It would be interesting to track the Google trend for searches on Chest Pain/Heart attack before and after the introduction of Vioxx as measure of pharmaco-vigilence.

*As part of the Clinical and Translational Science Awards, all Harvard Medical School affiliates must work together as single virtual unit, sharing data for clnical research. SHRINE is an innovative, web-services based federated data mining tool that enables clinical research among all the data at all Harvard hospitals with appropriate privacy protection and IRB oversight.

All of these new approaches go behind claims data to provide novel indicators that can be used to measure quality. I predict that all these novel sources of data will become increasingly important as stimulus funds become available and clinicians are incentivized based on quality, not quantity, of care delivered.