Overcoming the Film-based Mindset in a Technology-enabled System Through the Analysis and Conscious Design of Radiology Workflow
 
Authors:
Christopher R. Trimble, University of California, Irvine Medical Center; Ryan K. Olsen, MBA, MPH
 
Hypothesis:
A significant portion of lost revenue due to denial of insurance claims can be explained by inadequacies in the flow of data, due to suboptimal information workflow design.
 
Introduction:
Denial of insurance claims represents a major amount of lost professional revenue for the radiology department at the authors’ medical institution. Denied claims for studies performed for the emergency department specifically comprise approximately 20% of this lost revenue.

Currently, data population in the billing systems is automated via information feeds from the radiology information system (RIS) and hospital information system (HIS). The medical center is also on the cusp of a phased implementation of an enterprise-wide updated HIS and electronic medical record (EMR).

Suspecting that much of the lost revenue is due to inadequate information exchange between the information systems, the authors sought to quantify lost revenue and stratify by reason for denial for all denied claims for emergency radiology studies. Information flow within and between the RIS, HIS, and billing information systems was also examined.

 
Methods:
To examine lost emergency radiology professional revenue, a billing audit was performed from the beginning of the fiscal year 7/1/2009 to 9/1/2009. Indications for denial were grouped either according to missing/inconsistent patient data or miscellaneous. Indications for inclusion in former category included errors in demographic information and diagnosis/billing codes (ICD-9 and CPT).

To evaluate the information flow, a detailed process-oriented workflow analysis was performed, originating from the point of study requisition and ending with the submission of the billing claim. Data collection methods consisted of personal interviews with PACS and RIS managers, radiology technicians, network administrators, billing specialists, radiologists, and emergency medicine physicians. Detailed process flow diagrams were constructed with particular focus on the exchange of data between different information systems.

 
Results:
Threatened monthly professional revenue losses due to denied claims for missing or inconsistent clinical information were $29,530 for studies performed for the Emergency Department. Of the threatened revenue losses, 67% were recuperated through manual information reconciliation and claim resubmission. Despite efforts for recovery, 33% of the revenue was lost due to late penalties, irreconcilable clinical patient information, and inability to process all denied claims. With a monthly full time employee cost of $ 3,199 for the process of revenue recovery, total monthly losses amounted to $32,729, giving an estimated annual professional revenue loss due to denied claims for missing or inconsistent clinical information of $392,747.

Detailed process flow analysis revealed several relevant limitations in the flow of information between the different clinical systems including:

  • Unidirectional batch information transmission between systems;
  • The need of the billing system to reconcile information from both the HIS and the RIS;
  • Redundant manual data entry; and
  • Different coding schema between the RIS and billing systems.
Discussion:
The data support the authors’ suspicion that a significant amount of lost professional revenue is due to missing or inconsistent clinical information. Performing a detailed, process-oriented workflow analysis was valuable in identifying several inadequacies in the design of data flow between information systems that could account for a large portion of these billing errors.

Unidirectional and batch information transmission imposes the most limits on current workflow. It has serialized the flow of information—in essence creating three shadow systems that contain the same patient information, but must be updated separately. Once a study requisition has been made by the emergency physician and transmitted to the RIS system from the HIS, any subsequent changes to demographic or diagnosis information in the HIS would not be reflected in the RIS database. The billing system would then receive inconsistent information from the HIS and RIS systems, and the claim would be denied due to inconsistent information. Reconciliation of incomplete billing information is further complicated by the need to manually resolve this data between three separate information systems. Implementation of bidirectional HL7 interface engines would obviate the need for manual synchronization of information between the three databases.

Despite the possibilities of the automated flow of information between systems, our analysis also showed three steps of redundant manual data entry. Manual entry of diagnosis codes were completed either by the emergency physician during the study requisition (rarely), or by a tech during the time of study “arrival” in the RIS. The codes were again entered manually during the time of study completion by the radiology tech. Further, because diagnosis coding schemas were different in the RIS and billing systems, a translational database had to be kept and manually updated in order to assure proper data flow between the systems.

Besides the superfluous costs of the increased manual labor, a process involving so many steps of manual data entry not only carries additive risk of three potential sites of human error, but also carries an extra risk of systematizing errors due to the serial flow of information. Such errors not only affect financial reimbursement, but also can have negative effects on the quality of patient care. Elimination of duplicative data entry steps could be as easy as configuring autopopulation of codes in some instances. In other instances it is significantly limited by inherent system incompatibilities.

It seems that in the case of our institution, a lack of conscious design of technology-enabled radiology workflows has largely resulted in an automated film-based workflow. On the eve of the adoption of a new electronic medical record and planned upgrades to RIS and PACS, an audit of current information workflow between the clinical information systems has helped shift focus beyond inherent capabilities of the clinical information systems alone. Optimization of the integration and interoperability of clinical information systems in order to support seamless information flow clearly has benefits, both financial and in terms of quality of care.

 
Conclusion:
Poor engineering of information flow between systems can account for as much as $392,747 in lost annual professional revenue due to denied claims for missing or inconsistent clinical data for emergency radiology studies at the authors’ home institution. There are both quality of care and financial benefits in information and workflow analysis when adopting and integrating clinical information systems such as to assure dynamic information flow, interoperability, eliminate shadow systems, and reduce errors of duplicative data entry.
 
References:
Halsted MJ, Froehle CM. Design, Implementation, and Assessment of a Radiology Workflow Management System Am. J. Roentgenol. 2008;191:321-327

Siegel E, Reiner B, et al. Reengineering Workflow: The Radiologist’s Perspective. PACS. 2006;97-123.