NQMC - Expert Resources - Expert Commentary

Perspective
Electronic Health Records in a Large, Integrated Health System: It's Automatic....NOT! At Least, Not Yet.
By: Jed Weissberg, MD

One of the promises of the proposed investments in Health Information Technology is to more readily generate standardized quality measures. The following commentary captures the experience within Kaiser Permanente on this desirable outcome.

Push F2 for HEDIS, F3 for Joint Commission Core measures, F4 for ACOVE, F5 for quality measures yet to be defined...

Ah, were it only so. Even in these early days of electronic health record (EHR) implementation, we've graduated from 'unconscious incompetence' to 'conscious incompetence' in generating standardized quality measures from our routine EHR documentation. From the wished-for 'F2', we've now come face-to-face with the necessity to gain agreement and train clinicians on the 'proper' places for clinical documentation in our flexible EHRs, as well as with the diversity of data required to assess and drive quality. Many integrated health care organizations like ours are grappling with the most efficient ways to capture data for quality measures and all are still in 'exploratory' mode.

The Frontier of EHR Implementation

In 2003, Kaiser Permanente decided to purchase a suite of software applications from Epic Systems and deploy a program-wide electronic health record (EHR) system we termed Kaiser Permanente HealthConnect™ (KP HealthConnect). KP HealthConnect enables our clinicians and employees to manage the health care and administrative needs of our 8.6 million members across eight geographic regions in a completely integrated and seamless manner (1). Six years into implementation, where do we stand, and what have we found along the way with respect to using the EHR to measure and improve quality?

Most quality measures have grown up in an environment of quarterly or yearly retrospective reporting and analysis. However, if we wish to use any EHRs to help drive improvement of quality in the office visit setting, we need to use clinical decision support tools real-time with just the right sensitivity and specificity to make recommendations to clinicians appropriately but without inducing either alert fatigue or, worse yet, alert rage (2). In our initial attempts to utilize such computerized decision support, we found that available tools lacked flexibility in presentation and programming logic. Additionally, each new version of KP HealthConnect comes from the vendor with additional functionality that requires us to rethink previously implemented decision support and broadens the possibilities for new uses. For population-based approaches to QI, we need to abstract data from EHRs in order to form disease and device registries for purposes of outreach or safety monitoring. For many EHRs, such registries currently reside outside of the core EHR product, yet we need to be able to access and send information easily back and forth. Such work requires physician leadership and oversight, as well as technical expertise in IT and quality measurement.

At Kaiser Permanente, we are migrating from measurement activities using traditional ancillary systems and data sources to KP HealthConnect. But until that transition is complete, additional programming and analytic resources have had to be used for even routine reporting.

Life on the Leading Edge

There are some lessons and arrows pointing the way towards achieving a state of 'conscious competence.' One example is the leadership and collaboration shown by Kaiser Permanente's nursing staff across our national program, who decided at the outset of our inpatient implementation to agree to a single, standardized set of flow sheets for their documentation. This, paired with training on proper sites for data entry, allows us to automatically extract critical data elements, both for usual Joint Commission reporting, as well as for concurrent reports used for quality interventions during the index hospitalization. In the inpatient setting, the rich clinical data (vital signs and laboratory values) are enabling our researchers to develop algorithms to generate 'risk of deterioration' scores, which could be used to trigger versions of emergency response systems on the part of hospital personnel.

In the outpatient setting, we are learning how to meld traditional office encounter opportunities to improve care with innovative initiatives that harness the power of all clinical staff (not just those in primary care) to recommend care interventions to patients. For example, in our Southern California region we've implemented an approach termed the "Proactive Office Encounter," in which front office staff is supplied with relevant care gap information for members coming to clinic that day. Staff are then trained and empowered to call such needs to the members' attention and to facilitate scheduling, or otherwise getting the needed testing accomplished. This, plus the use of registry data to power outreach via letters, emails, and phone calls (live or computer-telephony), is helping us drive our HEDIS measure performance to our highest levels ever (3).

But beyond traditional HEDIS, what else can the EHR help us do? Medication safety is another hoped for benefit of EHRs and computerized physician order entry, but we've found that the usual drug-drug interaction pairings available from commercial vendors are far too sensitive, and the utility in our EHR doesn't allow us to use Boolean logic and other clinical data to help refine the hit rate. We've had to work with our drug-drug list vendor to 'promote' and 'demote' particular level 1 and level 2 pairs in order to have actual impact on safety. We are doing this for ourselves, just as every implementer of EHRs is doing across the country. Such duplication is inefficient and unnecessary.

Though quality measurement is often oriented to a particular disease or population, EHRs can also inform the care of individual patients. We are piloting a computer presentation of risk and recommended interventions for a cohort of our Hawaii members that goes beyond simply counting 'care gaps' for an individual or practice and presents a more holistic view of an individual's risk status. Using an off-line mathematical simulation tool (4), we are presenting a prioritized list of medical interventions to physicians and patients in an attempt to induce desired behavior change. Early results are encouraging. Such whole-person and near real-time presentations of data are impossible without the clinical data of an EHR.

How about continuity of care? This critically important dimension does not yet have widely accepted or standardized measures, yet the rich data in EHRs across an integrated system of care should allow us to test hypotheses and signals, such as the number of practitioners involved in a patient's care or the use of staff messaging functions concerning a patient. For end-of-life care, we also are in the process of standardizing the location within our EHR of such critical data as Advance Directives and Physician Orders for Life Sustaining Therapies to enable all caregivers to access patient wishes and joint treatment plans and goals. Such locations for care critical data should be clearly indicated and locatable in every EHR.

Health Information Exchange

What happens if our members are seen outside of our care settings or if they happen to move or change insurance status? KP HealthConnect enables members to view their clinical record, and we are piloting mechanisms for portability of their electronic data via thumb drives and Web-based personal health record systems (PHRs) such as Microsoft's Health Vault. We are also exchanging data for patient care in a Regional Health Information Organization in Colorado, and we are finalizing a limited production implementation as a node in the Nationwide Health Information Network to exchange data with the Veterans Health Administration for shared-care patients, but such data interoperability efforts are still early and require much organizational effort. The aggregation of individual quality data to population-based assessments will require such functions to operate seamlessly and transparently in the background. Strict adherence to standardized data and terminologies such as LOINC, HL7/CDA/CCD, SNOMED, and DICOM will be essential if information is to be easily shared across systems (5). Loosening the standards is one strategy proposed to ease the adoption of electronic records systems, but this has an adverse impact on the utility of data from these systems for QI both individually and collectively.

The Future

Kaiser Permanente made the investment of our members' dollars to buy and implement KP HealthConnect, an EHR that spans both ambulatory and hospital care settings. Six years in, we are beginning to be able to automatically generate usual quality measures and are using the data to drive clinical improvement on our way to being a health care system that fully embodies the IOM's six aims. We believe that such investments are essential to better health care for all Americans, yet we would very much like to see reduced duplication of effort by us and other leading systems implementing EHRs as we develop and use such tools. The following could aid in managing duplication and associated costs:

Collaboration amongst the software vendors, or via collaboratives sponsored by such bodies as AHRQ.
The NQMC can encourage HIT vendors to state how users would generate each measure listed in order to achieve the desired quality improvements more easily and quickly.
The certification process for EHRs could include standards for vendors to demonstrate an easy-to-implement approach to standardized measure generation.
The NQMC could also start listing measures derived from the clinically enriched datasets enabled by EHRs and document specifications sufficiently detailed for listing in AHRQ's USHIK. USHIK's large quantity of standardized data has become part of the national backbone of data interoperability. Standardized data elements specified by CHI, HITSP, HIPAA, CCHIT, and more are well documented. (6)
Editor's Note: NQMC plans to include measure specifications specific to EHRs (or other clinically enriched data sets) for any eligible measures submitted with this information.

We believe the recent stimulus monies in the American Recovery and Reinvestment Act would best be directed towards mature systems where it is possible to go beyond initial broad implementation to pushing the edge of EHR development and function so that, as we incent a more pervasive use of EHR technology, we are better able to 'start with the end in mind' and achieve the quality improvements that EHRs promise.

Author

Jed Weissberg, MD
Kaiser Foundation Health Plan and Hospitals, Inc., Oakland, California

Disclaimer

The views and opinions expressed are those of the author and do not necessarily state or reflect those of the National Quality Measures Clearinghouse™ (NQMC), the Agency for Healthcare Research and Quality (AHRQ), or its contractor ECRI Institute.

Potential Conflicts of Interest

Dr. Weissberg states no conflicts of interest.

References

Chen C, Garrido T, Chock D, Okawa G, Liang L. "The Kaiser Permanente Electronic Health Record: Transforming and Streamlining Modalities of Care," Health Affairs. 2009; 28 (2): 323-333.

Silvestre A, Sue VM, Allen JY. "If You Build It, Will They Come? The Kaiser Permanente Model of Online Health Care," Health Affairs. 2009; 28 (2): 334-344.


Isaac T, Weissman JS, Davis RB, Massagli M, Cyrulik A, Sands DZ, Weingart SN. "Overrides of Medication Alerts in Ambulatory Care," Archives of Internal Medicine. 2009; 169 (3): 303-311.


NCQA Quality Compass 2008.


Eddy DM, Schlessinger L. "Clinical Outcomes and Cost-Effectiveness of Strategies for Managing People at High Risk for Diabetes," Annals of Internal Medicine. 2005; 143 (4): 251-264.


Dolin RH, Mattison JE, Cohn S, Campbell KE, Wiesenthal AM, et al. "Kaiser Permanente's Convergent Medical Terminology," Stud Health Technol Inform. 2004; 107 (Pt 1): 346-50. PMID: 15360832 [PubMed - indexed for MEDLINE]


http://ushik.ahrq.gov/registry/index.html?Referer=Index

abrir aquí para acceder al documento AHRQ NQMC completo:
NQMC - Expert Resources - Expert Commentary