The Physician Coding Audit and Feedback System was developed for a renowned cancer diagnosis, treatment and research institution. As part of their clinical duties, physicians must choose billing codes for the Evaluation & Management services provided in their outpatient clinics (the typical patient visit to a doctor). These codes are required by Medicare and other insurers for reimbursement of services. There are three categories of office visits (New Patient, MD-Requested Consultation, and Follow-up), and each category has five different levels based on complexity and other factors for a total of 15 code choices in this setting. It is essential that the physicians follow strict Medicare regulations for choosing the correct code and that their documentation ("note") supports the code chosen. Inaccurate code selection and submission can expose the institution and physician to government investigation, negative publicity, financial penalties, and distraction from patient care.

Over the past 10-15 years, Medicare and the federal government have increased their efforts to investigate healthcare fraud and abuse. Since research has shown inappropriate reimbursement represents a huge dollar amount to the taxpayer, the government and public now demand full accountability of every Medicare dollar spent. Investigations are not limited to flagrant Medicare fraud cases, but also include errors due to lack of knowledge. In addition, the law makes a physician personally liable for improper coding, regardless of who selects the code.

In 2004, Christensen/Roberts Solutions was selected by the cancer center’s Compliance Officer to study their coding selection process in the outpatient clinic with the goal of improving accuracy. The following constraints that inhibited desired performance were presented to CRS:

• The complicated nature of the Medicare rules presented an inherent difficulty for the physicians to incorporate into their clinical routines,
• Medicare Rules are not clinically intuitive, are not medically driven and do not aid in patient care, leading to resistance by the physician’s population to learn the coding rules,
• Medicare Rules, in general, were created for non-medical personal and involve a 'counting' system for key components and artificial pseudo-estimates of the complexity of the visit which is considered insulting/non-medical to physicians,
• Physicians have extreme time constraints,
• Some physicians do not believe coding is their responsibility and lack basic understanding of the consequences of inaccurate coding.

The original request was to improve code selection accuracy through training. CRS asked for, and was granted, the opportunity to conduct its own independent analysis. CRS retained the services of a physician consultant who acted as the Subject Matter Expert and physician-advocate during the analysis and solution design.

Within six months, CRS presented its results. The analysis determined that:

• Most coding errors fall into one of 13 root causes (error groups) and the number of possible errors was finite.
• Most physicians tended to make the same coding errors repeatedly.
• Most coding errors can be corrected by a simple knowledge-based solution, eliminating the need for physicians to learn 90% of the Regulations that would not apply to their specific errors.
• The Medicare rules are confusing and often interpreted differently leading to inconsistent coding decisions.
• Physicians, in general, wanted direct, personalized code-selection feedback.
• Significant resistance should be anticipated in the implementation of any process that required an additional investment of time for physicians.

Upon acceptance of the analysis results, CRS presented the following recommendations:

1. Identify Regulations that cause variations in coding interpretation, come to a consensus on a single interpretation, and communicate the interpretation across the entire cancer center.
2. Eliminate the recommendation for training, which would have required putting all 400 cancer center physicians through generic, classroom training.
3. Create a consistent, reproducible, documentable audit process to analyze and audit the codes selected by the physicians to identify the errors at a very specific and individual level.
4. Test the new audit and feedback process via a rigorous “paper-based” pilot period before committing to technology.
5. Produce a database-driven audit and feedback tool to assist Coding Auditors in analyzing the accuracy of the code selected based on the chart documentation. The tool would identify errors and provide specific feedback, track overall errors by service and by physician across the institution, and quantify the errors in a dollar amount.
6. Convene a sponsor group, identify potential areas of resistance and determine policies to address and resolve legitimate resistance issues.

(Required)

• Supports performers through best practice processes.
• Establishes, or aids in establishing, goals.
• Minimizes terminology translation or interpretation.
• Provides access to supporting knowledge resources.
• Focuses on task(s), processes, and the natural flow of work.
• Reduces or eliminates the need for training/learning.
• Supports performance FIRST, and learning only as a secondary consequence of doing.
• Is innovative with respect to supporting performance.

1. Automate the process of auditing a physician’s code selection choice, using a set of established criteria for interpreting Medicare rules in a language familiar to the cancer center.
2. Automate the generation of the correct code using the auditor's answers to the interview questions employing a method by which the coding rules are converted into numerical representations in order for the computer to be able to analyze the process.
3. Identify specific coding errors and the root cause(s) of the error(s).
4. Provide specific, corrective feedback for each error through text-based explanations and recommendations for correcting the errors including references to job aids specific to the error.
5. Provide management utilities for tracking and reporting results.

Support Staff, whose role it is to select charts to be audited and assign those charts to the Coding Auditors.

1. The application has a workload feature that allows a support staff member to enter charts and assign them to a Coding Auditor. The application then allows the Coding Auditor to manage the audit workload based on their unique login.
2. The Chart Entry screen provides error checking to ensure that the support staff member selects the correct physician name and then the application automatically assigns a Physician's "Anonymous ID."

Coding Auditors, whose role it is to perform audits on each physician to determine the accuracy of their code selection.

1. The application generates the correct (documented) code through a simple interview interface in which the auditor answers questions (no need to know all the complicated key component levels and how to produce a final code from those components).
2. The application allows auditors to perform a partial audit and resume at any time. This functionality adapts to the auditor’s schedule which is frequently interrupted, and allows the auditor to find additional data if necessary and then re-open the audit.
3. The application allows auditors to change any answer. If an answer is changed, the application automatically resets the downstream and upstream questions affected by the change. The application also indicates what remaining questions need to be answered to finish the audit.
4. The audit process uses language common to all auditors.
5. Where known interpretation issues existed, pop-up screens and other cognitive aids are provided at the moment of need that explain the cancer center’s interpretation of the Regulation.
6. The application allows the Coding Auditors to view the text feedback that the Physician would receive in their final feedback report at any time during the audit.
7. The application is developed on a thin client using a standard browser so work can be accomplished on-site or off-site allowing for remote use.

Coding Managers, who oversee the consistency and accuracy of the Coding Auditors.

1. Once an audit is “complete” answers may be reviewed and changed as needed. This is a key function, as it allows for the establishment of a formal Quality Control process in which a Coding Manager is able to review and change any audited chart, preserving the original audited chart for comparison.
2. Easy printing of a Chart Summary for use in coaching to improve auditor’s use of the application.
3. Ability to add personalized comments allowing a certain degree of customization and perspective to the final reports presented to the physician.

Compliance Officers who provide the audit results and feedback to physicians, the service chiefs, and Hospital Administrators.

The application provides:

1. Individual Physician Reports, which track how often a Physician is making an error and expresses this in monetary terms compared over time.
2. Service Level reports, which compare coding accuracy progress over time to show overall organizational improvement and comparison against target metrics; differences among services regarding Root Cause can be easily assessed.
3. Physician Detail Reports, which summarize the financial impact of coding errors and improvements.
4. A Charge Reconciliation Report, which keeps track of corrected billable charges.

Physicians who receive detailed reports including direct, personalized feedback on how to improve their coding.

1. The application produces a Single Chart Summary report which is then attached to each annotated chart note outlining the coding issue and how to fix it eliminating the need for unproductive, time-consuming and costly training.

Sample screen

The example below shows the first screen where the auditors begin to answer questions in a logical progression. The data gathered during the "interview" allows the application to "calculate" the correct code based on the answers the Coding Auditor enters as the Auditor reviews the physician's documentation. The beginning questions (see the left Navigation Bar) address issues which could immediately identify the chart as Non-Billable and stops the audit, thus saving time/energy by eliminating the need to answer further questions.
The question “groups” outlined on the left navigation bar address the 13 root causes identified in the analysis in a logical YES/NO progression and probe to find the reasons these coding errors were made. As the reasons become clear, the application automatically inserts pre-designed comments that describe the problem, how to fix it, and where they can go for further support.

 

(Required)

This entry is (check one):    In production (being used today in a live work setting)         In a formative stage (prototype, proof-of-concept, introduced a sample of its intended users)                   

(Required)

1. Supports performers through best practice processes.

A major component of the design and development process was the creation of a single, consistent, institution-wide E&M outpatient coding and auditing process. E&M coding rules are based on confusing Guidelines published by Medicare that are subject to wide variations in interpretation and can lead to conflicting opinions/feedback. As a result, the project team underwent a rigorous, six-month effort to develop a unified, consistent, and defensible audit and feedback process (workflow) that represented the major underpinning of the Coding Audit and Feedback System. The audit and feedback process incorporated input from external and internal subject matter experts, industry best practices, and physician representatives to ensure language and style were appropriate for the physician and auditor audiences.

2. Establishes, or aids in establishing, goals.

Helps performers establish what they can do, want to do, or where to go, based on stored data/information.

Coding Auditors:

The goal of the application is to compare the code selected by the Physician with the code based on Medicare Guidelines. To accomplish this, the code is broken down into its required components: Chief Complaint (CC); History of Present Illness (HPI); Review of Systems (ROS); Past, Family and Social Histories (PFSH), History (Hx-Total), Physical Exam (PE); and Medical Decision Making (MDM-Total). By comparing the components of the code the Physician chose versus the correct/documented code, the application can identify the reasons, if any, the code was incorrect. The comparisons are made possible by converting the components into numbers. This method allows for easy computerized comparisons and a more objective approach to judging the documentation. The goal of the process, comparing the components of the code selected versus the code selected by Medicare Guidelines, is always visually apparent in the Single Chart Audit Table (SCAT) at the top right of the Audit Screen; the components “build” as the Auditor progresses through the interview.
Coding Managers:

The data collected via the question and answer interface can be used by the Coding Manager in quality control to identify performance issues. The data would allow the Coding Manager to determine exactly where the auditor’s reasoning went astray, thus aiding greatly in giving the auditors direct feedback.

Auditor workloads can easily be viewed via the middle section of the Desktop module labeled “Charts To Be Audited” which is searchable by auditor’s name. This allows for better workload balancing between auditors.

For Physicians and Hospital Administrators:

The goal of this program is the reduction of coding errors. The application has the ability to report current and prior coding error rates, broken down by Root Cause, Physician and Service. The reports generate the audit results in a quantifiable manner, allowing comparison among individual physicians and services (physician groups). Current vs. Prior results are clearly indicated.)

3. Minimizes terminology translation or interpretation.

All coding terms and language are normalized to the cancer center's standard language to reduce confusion between Medicare terminology and the cancer center-specific terminology.

Numerical translations are automated by the program.

4. Provides access to supporting and learning resources.

The application included rollover definitions and point of performance definitions /explanations bulleted to known areas of previous confusion.

Where more in depth process support is required, the application provides embedded tools and references at the point of need:

5.  Focuses on task(s), processes, and the natural flow of work.

The recommended task sequence (workflow) is represented in left Navigation Bar.

The sequence of questions (workflow) quickly identifies documentation omissions which would make the chart Non-Billable, decreasing unnecessary auditing time.

Intrastep workflow is built into the application through input logic; users are guided through the workflow using a simple interview-based question/answer model. Questions are progressively disclosed based on previous answers. Color coding of both left Navigation Bar Steps and Q/A allows for easy detection of what still needs to be completed by the auditor.

In addition, complete backwards-oriented logic is built into application, allowing users to retrace their steps to any point in the interview and resetting the logic state at every point.

6. Reduces or eliminates the need for training/learning.

For the Auditor: The built in interpretations, rollover definitions and code-generating logic decreases the need for auditor training.

For the Physician: The original strategy to achieve coding accuracy was to train each of the 400 physicians who provide this service to patients at the cancer center on the billing and coding rules based on the Medicare Guidelines. However, the performance analysis revealed that those physicians typically repeated the same errors, that the errors fell into one of 13 Root Causes, and that most errors could be eliminated by a short knowledge intervention of approximately 15 minutes. Therefore, with the ability to identify a physician’s individual coding errors and automatically provide specific/directed feedback, the cancer center could eliminate costly and ineffective generic training and reduce the disruption to patient care and research time.

Sample section from Physician Feedback Report.

7.  Supports performance FIRST, and learning only as a secondary consequence of doing.

For the Auditor: The structured, guided audit process allows the auditor to complete audits without the need to continually check Medicare rules and regulations. Auditors can concentrate on the major task, since the application performs the analysis and calculations to determine the appropriate codes.

The entire training period for Auditors to learn to use the application was a half-day knowledge transfer during which the Auditors were able to complete an audit by the end of the session. The time to complete an audit was significantly decreased, and the audit results generated much more usable information.

(Required)

(Required)

(Required)

If this is a prototype or proof-of-concept, what are the anticipated performance improvement results?  What formative data have been collected to-date?

The Coding Audit and Feedback System entered into full production on March 1, 2006.

During the initial rollout of the new system, a "high-touch" approach was followed. Auditors and Coding Managers attend each feedback session and initiate the session with a presentation that quickly reviews the importance of coding accurately and an introduction to the new audit system, and then reviews the Service Level Reports (with the results of the recent audit). Discussion ensues regarding reasons for the coding errors (if present). It is expected that after the original round of feedback sessions, the feedback reports generated by the system can be sent directly to the physicians, service chiefs, and hospital administrators, further reducing cost.

The initial rollout to the physicians and administrators has been well received. In the first five months after release, over 200 of the 400 cancer center physicians have been audited and results have been returned to the service chiefs and executive management. Personalized feedback generated by the application has been provided to approximately 75 physicians. By the quality and quantity of targeted questions asked during the feedback sessions, the physicians appear to understand the audit results and feedback, appear to be positively engaged, and agree that the personalized feedback can be put to use immediately. They have also indicated their appreciation for direct feedback as opposed to training, and have shown appreciation that the “solution” to the coding accuracy issue acknowledged (respected) the physician’s extreme time constraints.

It should be noted that a significant part of the initial success of the rollout of the application was due to careful attention to the "people side of change" and the incorporation of a formal change management process. It is our experience that even the best designed technology solutions often fail due to an underestimation and lack of preparation for all the resistance to come (a change in the way people work, especially when it impacts a large and varied audience such as the case in this project). Resistance, manifested in its different forms, was encountered throughout the design, development and implementation phases, but the Change Management preparation aided greatly in dealing with it. We would like to acknowledge the help of the folks at Prosci Corporation whose change management process was used.

Among the unanticipated positive outcomes of the project is the decision to hire an outsourcing company to fulfill the Coding auditors' function, reducing operating costs and allowing the cancer center Coding Auditors to serve the more important and higher-skilled function of Quality Assurance and providing feedback to physicians. This decision was made possible by the realization that the application, through its structured workflow and integrated Guideline interpretations, made it possible to use non-cancer center resources.

Coding Auditors have reported a significant increase in job satisfaction by being able to spend more time providing feedback to physicians. Auditors have indicated their satisfaction with the efficiency and accuracy of the application.

The Compliance Officer has already used the initial data generated by the application in executive-management presentations for policy-setting and resource allocation purposes.

The Auditing Manager has reported more effective and efficient coaching sessions with the Auditors using the printable Chart Summary and the Q/A functions.

Although most statistics are not available at the moment, the following data are presently being collected to measure business results:

• Time per audit
• # of audits performed
• Time to prepare and present the physician feedback (during initial roll-out)
• Number of physicians (and charts) audited per year
• Number of physicians provided feedback per year
• Number and change in physician coding errors over time
• Root Cause and change in physician coding error over time

Describe anything else that contributes to your submission being an exceptional and/or innovative PCD Solution entry.

Feedback from the cancer center Compliance Officer: “[Physician Coding Audit and Feedback System] is a powerful and easy-to-use application that elegantly reduces an unbearably complex process to something manageable and comprehensible.”