Increase in coding productivity. Early studies have shown that, for some coding applications, there is a net increase in coding speed. This is largely because the key point of CAC software is that codes have already been selected by the software and the coding professional need only review the preselected codes and make whatever changes are necessary. Certainly some medical domains lend themselves to higher speeds than others. Domains where the documentation tends to be highly repetitive (e.g., screening mammogram radiology reports) or where procedural techniques are fairly predictable (e.g., gastroenterology endoscopies) realize the greatest speeds.
Increase in coding consistency. It is difficult to make organizational improvements in coding when coding is done inconsistently from one day to the next or from one coder to the next. CAC is very consistent—even when it is not right. This consistency makes for a compelling case. Codes will be assigned in the same manner each time. In a structured input-based tool, codes linked to structured input are assigned once at the time the input is created. NLP rules-based software does not forget a rule one day and remember it another. Even mistakes would be consistently generated. Availability of coding audit trail. Because coding decisions made by CAC software are based on programming and on rules and statistical calculations, the reason a particular code was selected at any given time can be reconstructed and analyzed if necessary. System designers or developers may reconstruct such audit trails as needed.
Data query ability. The use of CAC data for such purposes as Joint Commission auditing, quality assurance measures, performance studies, credentialing, and research is an attractive feature of this technology. Many CAC systems offer different ways to query data from their systems, including prewritten “canned” reports, ad-hoc queries, and the use of structured query language (SQL) to access the data.
Potential for more comprehensive code assignment. In the case of outpatient claims, CMS-1500 forms have a limited number of fields for ICD-9-CM codes. Because of production demands, often only the codes necessary for reporting to third-party payers are captured. An advantage of CAC software is that it can code a report with ten diagnoses as quickly and easily as it can code a report with only two diagnoses. To the degree that a provider would allocate resources to capture all applicable codes, NLP could probably do it faster; this advantage simply falls under the category of productivity. But to the degree that a provider may elect to not be as thorough and simply report only the diagnoses required for reimbursement, the CAC advantage is that it could provide a more complete clinical picture, with all diagnoses successfully captured (or “recalled,” in computer science language), not merely the lucrative ones.
Potential increase in coding accuracy. Coding rules are a moving target, with clarifications offered quarterly in the case of ICD-9-CM and HCPCS Level II codes and monthly in the case of CPT codes. Revisions to the CPT, HCPCS, and ICD-9-CM code sets are made twice a year, and payer rules such as NCCI edits are frequently updated. Because of these changes, as well as the intricacy of coding rules, CAC software is possibly better equipped to code compliantly than even the most skilled coding professional. Furthermore, the axiom, “If it wasn’t documented, it wasn’t done,” may apply to CAC software better than to human coders. Both structured input-based and NLP-based CAC applications are incapable of assuming, leaping to conclusions, or even “reading between the lines”; the software tends to be more purely accurate (for better or worse), as it is based solely on the documentation. Potential decrease in coding costs.
Because CAC software does not need vacations or health insurance, it can code less expensively than human coders. It can work straight though lunch hour, and it can work in the middle of the night without a night-shift differential. By being available at times when humans are more expensive, turnaround time for code assignment can be shortened, resulting in improved accounts receivable management. As with any major change, the return on investment should take into consideration all relevant factors, including the initial investment in the system, how the tool is implemented, and ongoing costs.
