Computer assisted coding (CAC)

From Clinfowiki
Revision as of 18:35, 27 October 2014 by Yuan (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search


Computer assisted coding (CAC) in health care is an emerging technology or computerized tool that automates the coding process (such as ICD-9, and CPT) by generating a set of medical codes for review based on clinical documentation provided by health care professionals.

CAC uses structured input or natural language processing (NLP). NLP is software technology that uses artificial intelligence to extract data and information from a text form and converts it to medical codes. The software used is not dependent on a fully functioning electronic health record (EHR) but as adoption of EHRs increase and organizations implement ICD-10 it is expected that implementation of CAC will also increase. Using a NLP engine, CAC is able to scan clinical documentation and identify key terminology to suggest appropriate diagnostic and procedural codes. [1]


With the use of EHRs facilities are seeing a greater amount of documentation by clinicians that coders then have to read through. There is financial pressure to get claims to insurance companies as efficiently as possible. ICD-10 will also change the volume and structure of coding allowing for a higher level of specificity. Based on Canada’s experience with the transition to ICD-10, the initial coding productivity is anticipated to decline by nearly 50 percent. There are mainly two reasons for the decline. First, the number of diagnostic codes increases from 14567 to 69833 and inpatient procedure codes increase from 4000 to 71918. Second, with the increased granularity of code selections, the specificity of clinicians’ documentation follows. This will significantly affect coders’ productivity.[2] For these reasons, there is a need to improve the efficiency and productivity of coding processes. CAC also allows for an easier coding audit trail. By streamlining workflows, coders then have more time to perform more complex tasks. CAC has become a solution for many healthcare organizations in hope of mitigating some of the anticipated decrease in productivity.

CAC will not eliminate coding jobs as some have feared, but it can reduce the amount of hours spent on manually assigning codes. It is expected that in the future, coders will still need to review, validate and interpret coded data and will have opportunities to get involved in more knowledge based tasks such as retrieval and analysis of aggregate data, participation with process improvement teams, and providing input on coding guidelines to name a few. CAC remains an effective tool to assist coders by scanning documentations, notifying relevant information and suggesting codes. It is not to replace them as the decision maker. In other words, coders are transformed into coding auditors. Not only job losses are not expected, but in fact additional work is anticipated with the implementation of ICD-10. [1]


A research conducted by the American Health Information Management Association (AHIMA) and the Cleveland Clinic revealed the effectiveness of CAC technology. [3] Three major findings are as follow:

1. CAC, paired with coders, reduces coding time. Coders who used CAC to code inpatient records resulted in 22 percent reduction in time per record.

2. CAC, paired with coders, does not reduce accuracy. Having a 95 percent accuracy rate as a standard, study found coders were able to reduce the time to code without decreasing quality for both procedures and diagnoses. The accuracy was measured by Recall and Precision.

3. CAC tuning improves recall, precision over time. The tuning (continuous learning) of the NLP engine allowed the CAC recall rate to improve over time for diagnostic and procedural coding.


CAC depends on the amount of electronic records and its availability. CAC can only read electronic documentations. Paper format will not be benefited from CAC. Also, accuracy and efficiency will not improve with such technology if the level of documentation is absent. [1]

Optimization of Use

Even a sophisticated CAC cannot reach its maximal potential without the integrity of inputted data. A clinical documentation program is essential as clinical support to enhance the validity of code assignments. It is suggested healthcare organizations should have a strong clinical documentation program in place before implementing CAC. Concurrent coding should accompany CAC rather than retrospective coding to optimize coding efficiency. CAC allows two disciplines to work in a real time collaborative way while clinical documentation specialists validate codes by content and coders ensure guideline and payment compliance.[2]


An optimal CAC software coupled with a robust clinical documentation program should benefit health organizations in the following ways[4]:

1. Mitigate the risk of productivity loss

2. Accelerate claim submissions and billing processes

3. Improve cash flow and decrease accounts receivable days

4. Identify issues in clinical documentation

5. Facilitate cross-departmental communication and share best practices

6. Improve coding compliance


  1. Rollins, Genna. "Lean Coding Machine: Facilities Target Productivity and Job Satisfaction with Coding Automation." Journal of AHIMA 81, no.7 (July 2010) [1]
  2. AHIMA e-HIMTM Work Group on Computer-Assisted Coding. "Delving into Computer-assisted Coding" (AHIMA Practice Brief). Journal of AHIMA 75, no.10 (Nov-Dec 2004) [2]

Submitted by Sally Gibson


  1. 1.0 1.1 1.2 Crawford, M. (2013). Truth about Computer-Assisted Coding: A Consultant, HIM Professional, and Vendor Weigh in on the Real CAC Impact. Journal of AHIMA, 84, 24-27.
  2. 2.0 2.1 Tully, M., & Carmichael A. (2012). Computer-assisted coding and clinical documentation first things first. Healthcare financial management, 46-49.
  3. Dougherty, M., Seabold, S., & White, S. E. (2013). Study reveals hard facts on CAC. Journal of AHIMA, 84, 54-56.
  4. Cassidy, B. (2013). Ten more questions for CAC vendors. Journal of AHIMA. Retrieved from

Submitted by Yuanye Lu