Will accountable care organizations follow the lead of HIEs in analyzing data across participating providers, or surpass them?

In this series, we are examining ways health IT can best support the goals of Accountable Care Organizations (ACOs) for health reform.

In our first article we focused on how clinical care and administrative data, as well as software tools, can be arrayed to support quality and efficiency analytics and reporting for an ACO. Data analytic technology usually operates retrospectively on non-transactional data that can be accumulated from diverse systems. As such, while not easy it is perhaps the easiest part of architecting an ACO technical infrastructure. There are analytic challenges in accumulating, normalizing, linking and processing the "iceberg" of both data visible in measures and the greater quantities of less-visible, supportive data needed for analytics and reporting.

Cross-organizational analysis
There are also many political issues related to bringing together multi-organizational, constituent data, but the business risk being imposed on ACOs and the opportunity that comes from bringing in claims data from multiple payers should help pressure for pan-ACO cooperation in this area. There are also clear reasons for being able to analyze data across multiple ACO constituent organizations. Some very useful quality measures, for example, relate to readmission factors and other elements of multi-organizational care. It is not clear that reporting or analysis from individual EHRs will ever be able to easily access these measures without some sort of intermediate organization whether that is the ACO or not.

The challenges of improving, rather than just reporting on, quality and efficiency, however, are even more daunting. Quality and efficiency "management" rather than just "measurement" places a heavier reliance on ACO-wide sharing of transactional clinical and cost data and on workflow processes rather than just retrospective analysis and reporting. Some of the kinds of quality and efficiency management that ACOs will seek can certainly come from taking analytic learnings and manually applying them though programs and policies that induce changes in care. But in many respects manual application misses what many think can provide the greatest gains from using EHRs and electronic technology.

Many believe that bringing clinical record data and derived knowledge together electronically in the clinical workflow offers the best opportunities to make substantive impacts on quality and efficiency. Some of these activities fall under the broad and somewhat ill-defined title of "clinical decision support" (CDS) and medical error reduction, but some are also more mundane information sharing patterns that relate to where and how clinical record data can be accessed – by people and by systems – and what an EHR is for the broad community of an ACO.

Here is a concrete way of helping to visualize these issues and capabilities. Some communities now have what amounts to a single, shared EHR for all of their community and hospital providers. They are not just sharing a single brand of EHR that is implemented several times in different organizations. They actually have a single instance of an EHR where patient record data are seamlessly accessible in both community and hospital settings. At times, they are getting to this state by having all of the providers owned by the same care organization. At other times, they are getting there because the hospital(s) purchased a particular product and is aggressively providing EHR services to community physicians from the same data repository. In these environments, at the most basic level, the ACO organizational boundaries match the single instance EHR organizational boundaries. Said another way – there is a single, "community EHR" where community and inpatient providers can retrieve store, retrieve and manage patient records that include entries from multiple providers.

Such a community EHR offers great opportunities for information sharing – both viewing and electronic processing – with minimal interoperability needs. As such, it readily provides some functions that can be used to help manage quality and efficiency of care. It provides an infrastructure that supports many medical home concepts without having to work though all of the politics and workflow incentives for insuring that the right primary care provider has all of the information that is needed. It eases transition of care issues. Medication reconciliation when a patient moves into an inpatient setting can be done one time for each involved organization instead of each time a patient moves in and out of inpatient care. Patient record sharing is made technically simple because instead of having to achieve machine processable, semantic interoperability between multiple EHRs, authorization to query the record from the shared store is all that is necessary. Processes can be setup to automatically determine if a newly planned test has already been done. And the growing category of "managed" vs. "transactional" medical record data (summary record, problem list, medication list, allergy list, care plan) is much easier to share and support for all participants in a patient's care. A shared instance of some of these managed clinical record data eases processable information reconciliation between organizations, minimizes duplication of the work of managing these data, and provides a common platform for the trigger data that are the jumping off point for leveraging automated CDS and other quality and efficiency processes though electronic query and processing.

Most developing ACOs will not have a community-wide EHR and will, instead, have numerous different EHRs. These organizations will then need to try to emulate some of these information sharing capabilities and functions by combining the EHRs with private health information exchanges (HIEs).

Without stronger standards, however, it remains to be seen whether the private HIE products will be able to sufficiently overcome the interoperability challenges and provide similar functionality. If not, the difference between these models may become not only an indicator of ACO success, but an important component in the viability of independent community providers in an increasingly electronic clinical care world.