Computer-Assisted Coding at the Cleveland Clinic: A Strategic Solution: Addressing Clinical Documentation Improvement, ICD-10-CM/PCS Implementation, and More

By Kathy Hartman, RN, MSN, CNS; Shannon Connor Phillips, MD, MPH, FAAP; and Lyman Sornberger

How one organization is improving documentation and quality of care through innovation and transformation.

Healthcare reform, electronic health records (EHRs), pay for performance, and ICD-10-CM/PCS are just some of the influences changing the landscape of health information management (HIM) quality and finance. Administrative data is used to measure patient care and clinical performance for hospitals and providers. In order for the administrative and coded data to be valid, the care provided to the patient needs to be documented with accuracy and completeness. Documentation requirements will increase with the conversion to ICD-10-CM/PCS.

The burning platform of healthcare reform, ICD-10-CM/PCS, coding shortages, and the resulting increasing demands set the stage for alternative solutions. EHRs and computer-assisted coding (CAC) can be leveraged to automate and improve documentation, coding, data extraction, and ultimately patient care. At the Cleveland Clinic Health System (CCHS), we view this changing landscape as an opportunity for innovation and transformation.

CCHS consists of 10 hospitals. The main campus-the focus of this article-has approximately 1,300 beds, about 2,800 employed physicians and scientists, and about 53,000 acute care admissions, with a case mix index 2.33, length of stay 6.6 days.

Required publicly recorded quality measures at the main campus showed our performance trailing similar academic centers. At first glance, this data did not seem to accurately reflect the complexity of patients' medical conditions or care provided. In addition, patient safety indicator (PSI) results did not accurately match true patient outcomes. For that reason, a call to action was initiated.

PSI Project – 1 Year Retrospective Review

Hartman 1

Key CCHS stakeholders formed an enterprise committee that aligned key leaders in clinical care, finance, and quality. Specifically, leaders from finance, HIM, care management, quality, medical staff, and clinical institutes became a steering committee charged with improving our performance. Named the Documentation Extraction Reporting and Transformation (DERT) committee, this team has shaped the mission and vision of an innovative model to transform documentation and extraction practices to positively impact quality patient care, data integrity, documentation, clinical coding, and financial performance.

How are clinical outcomes generated? To answer this question, a discovery assessment was conducted consisting of a retrospective review of 880 cases for six discrete PSIs (3, 6, 7, 9, 11, and 12)-for specific PSI definitions, visit the Agency for Healthcare Research and Quality Web site at Of the 880 cases, 19 percent were identified as having documentation/coding opportunity but little could be done retrospectively. The opportunities identified during the review were categorized into clinical care, documentation, extraction, and reporting operational workgroups.

Outcomes = Data

Hartman 1

The CCHS culture establishes that all employees are caregivers. One of the lessons our organization learned, specifically for the HIM professionals, is that they are the frontline of quality and safety relating to reimbursement and reputation (see "Outcomes = Data" on page 25). As caregivers, we eagerly accept the role of stewards of data integrity.

The advent of ICD-10-CM/PCS was an additional driver to move quickly, as incomplete or inaccurate documentation would be evident immediately. The selection process for a CAC product involved initial requests for information, followed by product demonstrations by the top four identified vendors to the HIM selection committee. The members of the committee included stakeholders from the HIM (operations, coding, CDI, and EHR), IT, and finance departments. The team evaluated features, functionality, ability to interface into the current CCHS EHR, cost, and return on investment in order to make a recommendation.

A capital budget request was submitted to purchase the selected product. The project scope covered the CCHS main campus technical coding of inpatient, ER, ambulatory surgery, and ancillary cases in year one. A return on investment was calculated using coder productivity metrics.

Some identified issues that impact quality, data, and financial performance are:

  • Incomplete documentation
  • Incomplete or incorrect coding
  • Sequencing of coded data
  • Understanding the complexity of cases regarding both severity and risk of mortality
  • Data integrity/mapping

One of the most significant barriers to the documentation translation to codes is the language barrier between provider documentation and coded data.

Barriers and considerations regarding clinical documentation translation to coded data are:

  • EHR copy/paste function/large volume of progress notes
  • Importing diagnoses from outpatient problem lists to the inpatient episode of care
  • Templates
  • Missing documentation of all present on admission conditions
  • Lack of specificity

Recognizing this barrier, the DERT committee recommended a concurrent review process that included a concurrent query process specific to PSIs and a process to educate physicians and stakeholders. The team recognized one significant clinical documentation challenge-providers and coders often speak a different language.

One of the most important questions was: how does CCHS leverage CAC to address these issues?

The Clinical Documentation Challenge


ICD-10-CM/PCS Impact on Data Quality and Reporting

Readiness is the key. A strategic approach was used in the transition plan for ICD-10-CM/PCS implementation. By performing a comprehensive system assessment, we were able to evaluate the impact. Some of the findings:

  1. Ninety-one percent of the selected records have at least one code that is a "one to many" code in ICD-10-CM/PCS that today currently drives the DRG payment. This means it is either in the principal slot, a complication or comorbidity (CC), major CC (MCC), or a procedure that is used in calculating the correct DRG today (and likely under ICD-10-CM/PCS as well).
  2. We evaluated how many codes of the "one to many" are diagnosis codes, how many are procedure codes, and the number of claims where a diagnosis and procedure exist on the same claim that have "one to many" translations. The conclusion was that main campus dealt with more "one to many" codes than most facilities.
  3. In the record, audit documentation did exist in many of the records to support the specificity needed on the diagnosis portion of the claims, but the documentation specificity required to report procedures does not exist in most of the records we reviewed. Therefore, our conclusion was that we will need to educate physicians and query for additional documentation to support both ICD-10-CM/PCS, but our focus will primarily be on the PCS.
  4. We reviewed our top five specialties (cardiothoracic surgery, cardiology, medicine, orthopedics, and surgery) and identified focus areas in each service line.

Results – First Four Months


Using CAC with CDI

What is CAC? The tool uses current documentation in the EHR, which allows annotation of the documentation and assignment of codes based on the documentation through natural language processing (NLP). The "search and compare" functions assist the coders and CDI staff in reviewing vast amounts of data in a more efficient manner; query edits effectively, increase query rates, and improve MCC/CC capture rates; auto-suggested codes further improve productivity and accuracy; and word search functions allow CDI staff to search for terms that should be documented based on clinical data. Additionally, the document highlights what is new in each subsequent progress note and the ICD-10-CM/PCS-CM crosswalk and tips allow CDI staff to begin educating physicians about additional specificity needed for accurate coding.

CAC automates the process of clinical documentation review and generates codes for validation, providing benefits such as:

  • Reduced contract coding FTEs
  • Reduced accounts receivable days and "discharged not final billed" accounts
  • Improved capture of patient severity
  • Facilitated identification of PSI and hospital-acquired conditions (HACs)
  • Integration with the CDI program and improved working DRG accuracy and potential queries to physicians
  • Improved coder and CDI staff satisfaction
  • Additional support for ICD-10 readiness

The current CDI program will be utilized for physician education and adoption of ICD-10-CM/PCS. Successful adoption will require integration between HIM, CDI, case management, and physicians focused on improving physician documentation standards that result in protecting and improving CMI throughout the ICD-10-CM/PCS adoption period.

Two budget scenarios were produced in preparation for ICD-10-CM/PCS. In the conservative approach, it was assumed that only the main campus and Weston community hospitals would be live on CAC and would experience a 30 percent productivity improvement. A temporary 50 percent productivity decrease is projected for the ICD-10 implementation, but this is offset by an assumed 30 percent productivity improvement for sites on CAC. The net impact is a productivity decline of only 20 percent for these sites. Therefore, the budget for main campus and Weston was based on a 20 percent coder productivity loss requiring augmentation. All other facilities assume a 50 percent coder productivity loss for nine months. Labor augmentation was calculated accordingly.

In the aggressive approach, the assumption was made that all facilities would be live on CAC. Therefore, all sites would experience a net 20 percent coder productivity loss requiring staff augmentation.

The conservative approach estimates $11.9 million in staff augmentation while the aggressive approach estimates $5.5 million in staff augmentation, resulting in a difference of $6.3 million.

Main campus implemented CAC annotation in September 2011 for coding and October 2011 for CDI. NLP highlights everything in the patient record that is a diagnosis or procedure and automatically codes as far down the coding decision tree as possible. The auto suggestion functionality is scheduled to go live in the second quarter of 2012. The auto-suggestion coding allows the coder to review and select the code directly from the "suggested" terms or group of terms in a sentence.

Results to Date

  1. DERT observed a 69 percent decrease in expected versus observed PSIs (see "Results: First Four Months"). The decrease is a result of increased institutional awareness of PSIs/HACs, education for providers and coding staff and resources for just in time education, piloted concurrent review, clinical care improvement, and a process to check each case before it is finalized.
  2. Coding: Coders' six-month CAC satisfaction results indicate an improvement in user satisfaction from the baseline and a perceived improved productivity rate. The search and compare feature is the most appreciated feature, while searching for missing documents was the most challenging.
  3. CDI: To date, the results of CAC for CDI is a nine percent improved query rate and an 11 percent improved CC capture rate, while maintaining an approximate 70 percent physician agree rate. Employee satisfaction has also improved, and users identify the best feature of the product as the highlighting of terms.
  4. Post implementation: Auto-suggestion metrics will be obtained and will include productivity measurement.

CAC as Part of the Plan

Intelligence-driven documentation and technology tools such as CAC were an integral part of the overall plan to address both the transition to ICD-10-CM/PCS and the need for improved documentation and data integrity.

Further related enhancements and innovations planned for CCHS are expansion of the use of the CAC product to include inpatient professional coding, expansion of CAC to the community hospitals, and leveraging of CAC to improve EHR quality related to problem lists, copy/paste functions, and templates. In addition, CCHS will use predictive modeling for concurrent identification of PSI/HACs, LOS, and risk of mortality indicators to continually improve the quality of patient care and its translation through coding.

Kathy Hartman ( is senior director, health information management, Dr. Shannon Connor Phillips is quality and patient safety officer, and Lyman Sornberger is executive director, revenue cycle management at Cleveland Clinic Health System in Cleveland, OH.


The article "Putting the HIM in IM" in the May Journal of AHIMA incorrectly described a CPT code for acupuncture. The description of code 97811 should have read "Each additional 15 minutes of personal one-on-one contact with the patient, with reinsertion of needles." The Journal regrets the error.


Article citation:
Hartman, Kathy; Phillips, Shannon Connor; Sornberger, Lyman. "Computer-Assisted Coding at the Cleveland Clinic: A Strategic Solution: Addressing Clinical Documentation Improvement, ICD-10-CM/PCS Implementation, and More." Journal of AHIMA 83, no.7 (July 2012): 24-28.

Article citation:
Hartman, Kathy; Phillips, Shannon Connor; Sornberger, Lyman. "Computer-Assisted Coding at the Cleveland Clinic: A Strategic Solution: Addressing Clinical Documentation Improvement, ICD-10-CM/PCS Implementation, and More" Journal of AHIMA 83, no.7 (July 2012): 24-28.