This qualitative research study was undertaken in an effort to better understand how coding professionals feel about the use of computers and automated coding. Observation of the coding process in the hospital setting and interviews of coding professionals were conducted. The resultant data was analyzed using the grounded theory approach. Scrutiny-based data analysis techniques as well as processing techniques of 1,462 lines of data revealed 18 specific themes in four categories. Results indicate that coding professionals generally accept the use of computers in the coding workflow. There are strong emotions about automated coding, many of them negative. Much of the anxiety surrounding automated coding is related to an underlying mistrust and concern that decision makers will misuse, or overuse, the technology. Outpatient coders are more likely than inpatient coders to feel threatened by automated coding.
The manual workflow for capturing medical codes for reporting and claims processing is described as expensive and inefficient.1 Despite this, and despite the fact that multiple studies on the accuracy of automated coding appear promising,2–7 widespread adoption of automated coding has not yet occurred.8 In an attempt to learn about coding professionals’ attitudes toward automated coding, AHIMA conducted a survey of coding professionals in 2005. This survey revealed that coders are beginning to use automated coding software, and in general, coders are only slightly concerned that jobs will be eliminated with the use of automated coding.9
This qualitative research study was undertaken to investigate more fully how coding professionals feel about automated coding and the use of computers in the coding process. The purpose of this study was to determine how readily coding professionals accept the use of computers in the coding workflow and whether or not they feel threatened by automation.
The grounded theory approach was used to allow themes to emerge from the data that were gathered. The primary qualitative research techniques were personal interviews and field observation. A typical (nonautomated) coding process in a local hospital was observed. Personal interviews were conducted with coding professionals who do not use automated coding as well as with a coding professional who uses automated coding.
The study population consisted of coding professionals performing medical coding in hospitals for the purpose of reporting diagnoses and procedures on healthcare claims. Three semi-structured interviews of one hour duration were conducted. Interview subjects included coding professionals who do not use automated coding as well as those who are presently using such a system. They were all coding professionals who perform medical coding of hospital charts for inpatient and/or outpatient visits.
Field observation was conducted in the coding department of a local hospital that reflected the typical coding process employed today. Observation focused on how the coding is performed, including how much the coders use a computer and for what, as well as how often they use paper charts, a coding manual of any sort, or other medical references (medical dictionary, drug handbook, and so forth) and for what.
Data Gathering and Analysis
Recorded interviews were transcribed, and jottings during observation were expanded into detailed, typed field notes. This yielded a total of 1,462 lines of data in 36 typed pages. These data were edited with a grounded hermeneutic approach by reading the whole to begin to get a sense of emerging themes and concepts without overlaying presumptions.
Data analysis was initially done via scrutiny-based techniques. This revealed multiple potential themes and subthemes that were analyzed for similarities and differences. Subsequently, cutting and sorting techniques as well as limited use of key words in context (KWIC)10 were employed to narrow the multiple subthemes down to primary themes grouped into related main categories. These themes and categories were continually reviewed for consistency with the source data in an iterative manner. Resultant themes that emerged are described in this paper.
Four main categories of themes were identified with a total of 18 specific themes. Results for each theme are summarized here. Representative quotes from transcripts are also presented for each theme in Table 1.
Use of Encoders
Notably, all of the coding professionals involved in this study use encoder applications. Encoders are viewed as an essential tool to assist coding professionals today. Coders find it particularly useful when encoders provide “ready access to reference materials and alerts of various coding rules.” There is recognition, however, that encoders are just a tool and the coding professional must have a thorough knowledge of the coding classification system. “If the coder is not book savvy to begin with, if they’ve been taught how to code by the use of an encoder, rather than understanding the format of the coding book and the coding rules that are included in the coding book, they may very well arrive at the wrong code.”
Increased Complexity of Computer Records
Coding professionals relayed that they experienced increased complexity in the coding process when computerized records were employed. Hybrid records require that the coder work from both a paper record and computerized images, and the coder is continually moving back and forth from the computer screen to paper on the desk surface. In addition, shortcuts from familiarity with colored forms and unique formats may be lost. “The hybrid record causes a problem for coding in that everything is black and white. So in the past the efficiencies they had from knowing to go to the pink sheet is not there anymore because the color is not there. So, unless it’s specifically indexed as a document type, that complicates the coding process.”
The use of electronic health records (EHRs) poses a unique challenge for coding professionals. The clinical data analysis required to accurately and completely code a patient record is much more challenging in an EHR. “In an EHR the story is lost. In a patient record you have a story and you just follow along in the text of the story and you can see what has happened. In an EHR the story gets segregated by the entering author and that segregation doesn’t methodically flow.”
Coding professionals further relayed that these complexities slow down the overall coding process. “I could meet the ER [emergency room] coding quota easily and do more sitting there with the actual paper records. But now the computer can be slow especially with all the pages that an ER record might have.”
Use of Automated Coding
Coders recognize the benefits of automated coding systems. Interview subjects pointed out that it may result in significant gains in productivity (“They showed how fast it was and I thought that was really cool”) and potential gains in accuracy as well (“It could help us find a code that we might otherwise have missed”).
Anticipated effects on their job varied widely between inpatient and strictly outpatient coders. Inpatient coders appear to focus more on the productivity gains (“It will make the job go faster, so you can code twice as many cases”) but question when it will actually be suitable for use in production (“For use in more complex records, like inpatient records, it isn’t gonna be there for a long time”), whereas outpatient coders see the technology as more imminent and are more likely to wonder if there will be a job for them five years from now. “I would say 80 percent of the cases that I code are straightforward enough for the computer to be able to code correctly. Down the line will there actually be a need for me?”
Both inpatient and outpatient coders, however, indicated that automated coding will not entirely replace coding professionals. “I don’t think the coder is ever going to go away. There is only so much of our job that can be automated.” And both groups also indicated that use of automated coding systems will require coding professionals to increase their coding skills: “As you get more technology in here, coders will get harder and harder charts.” “The computer will do the easy stuff.”
Coders’ Emotions toward Automated Coding
Coding professionals express strong emotions on the subject of automated coding. For the most part, these emotions are negative. There is some wonder and amazement in response to the technology, but on a personal level the potential impact of the technology causes anger, anxiety, insecurity, mistrust, uncertainty, and worry. Notably, the mistrust seems to be focused on upper management and their lack of understanding of the complexity of the coder’s job with anticipated resultant misuse of the technology. “Where I used to work, the financial people, they totally didn’t get it. They just look at the AR and wonder what those coders are doing. They think it’s just a matter of looking up numbers. There is so much more to it, but they don’t know that. I think if they thought a computer could do our job, they’d replace us. Man, what a mess that would be.”
As noted above, the theme regarding the use of automated coding showed significant variance between inpatient and outpatient coders. Inpatient coders display more solid confidence that a computer cannot replace them. “A lot of coders worry that it will displace the coders but I could see where it did not displace the coders, it just made the coding job easier, it was just another tool that made it easier.” The key words in context (KWIC) technique revealed no words denoting negative emotions in the inpatient coder interview. In contrast, strictly outpatient coders used multiple words denoting negative emotions.
There are no data in this study to suggest that coding professionals are generally fearful of change or advances in technology. This study confirms that coding professionals accept automation. Use of computers, such as encoders, to aid the coding process is standard operating procedure. Computers are viewed as simply another tool to help the coding professional perform his or her job. An encoder, for example, is a tool that must be used properly and judiciously by the coding professional in order to assign codes correctly.
However, advances in technology that are useful for some segments of healthcare may not necessarily be an advantage to coding professionals. Electronic health records, whether fully electronic or in a hybrid state, have added complexity to the coding process. The lack of data-content standards is particularly problematic. In general, it takes longer to access the clinical information required for code assignment when using either a hybrid record or an EHR.
Automated coding is still an unknown entity, and coding professionals have strong emotions about this technology. Rogers’s theory on the diffusion of innovations states that innovation adoption is a process of uncertainty reduction. As individuals synthesize information about the technology, they form beliefs about the technology that drive their decision to adopt or not adopt it.11 Many coding professionals are still uncertain about automated coding.
A new theme emerged from this study by comparing interview transcripts from inpatient and outpatient coding professionals. Outpatient coders feel more threatened by automated coding than inpatient coders. This is no doubt validated by studies that report success with automated coding in outpatient domains but continued challenges in the inpatient domain.12, 13
Another important theme that emerged from this study is the potential mistrust of decision makers and concern that automated coding solutions will be misused. This is perhaps fueled by the fact that the complexities of the coding professionals’ job are difficult to articulate. One interview subject stated that “the difficulty of the coding, the rules and so on, is the hardest thing to explain to people, even people in the healthcare profession.” There are many instances where a judgment call is needed to consistently apply the underlying intent of coding guidelines and reporting requirements within the context of a particular patient visit.
The results of this study indicate that coding professionals generally accept the use of computers in the coding workflow. In regard specifically to automated coding solutions, outpatient coders are more likely to feel threatened than inpatient coders.
Much of the anxiety surrounding automated coding is related to an underlying mistrust of decision makers and the concern that they will misuse, or overuse, the technology.
Mary H. Stanfill, RHIA, CCS, CCS-P, is vice president of the professional practice resources division of AHIMA in Chicago, IL.
This work was completed as part of the required coursework for the BMI 526 Qualitative Research methods course at Oregon Health and Science University. Thanks to instructors Joan Ash and Ken Guappone for their expertise and insights in study design, particularly design of interview questions. Special thanks to the participants in the study who took time out of their busy work day to inform this work.
- AHIMA eHIM Workgroup on Computer-Assisted Coding. “Delving into Computer-assisted Coding.” Journal of AHIMA 75, no. 10 (2004, November–December): 48A–H. Available from the AHIMA Web site at http://library.ahima.org/xpedio/groups/public/documents/ahima/pub_bok1_025099.html.
- Hripcsak, G., C. Friedman, P. O. Alderson, W. DuMouchel, S. B. Johnson, and P. D. Clayton. “Unlocking Clinical Data from Narrative Reports.” Annals of Internal Medicine 122 (1995): 681–688.
- Elkins, J. S., C. Friedman, B. Boden-Albala, R. L. Sacco, and G. Hripcsak. “Coding Neuroradiology Reports for the Northern Manhattan Stroke Study: A Comparison of Natural Language Processing and Manual Review.” Computers and Biomedical Research 33 (2000): 1–10.
- Morris, W. C., D. T. Heinze, H. Warner, A. Primack, A. E. Morsch, R. E. Sheffer, et al. “Assessing the Accuracy of an Automated Coding System in Emergency Medicine.” Proceedings of the AMIA Annual Symposium (2000): 595–599.
- Hripcsak, G., J. M. Austin, P. O. Alderson, and C. Friedman. “Use of Natural Language Processing to Translate Clinical Information from a Database of 889,921 Chest Radiographic Reports.” Radiology 224, no. 1 (2002): 157–163.
- Mamlin, B. W., D. T. Heinze, and C. J. McDonald. “Automated Extraction and Normalization of Findings from Cancer-related Free-Text Radiology Reports.” Proceedings of the AMIA Annual Symposium (2003): 420–424.
- Friedman, C., L. Shagina, Y. Lussier, and G. Hripcsak. “Automated Encoding of Clinical Documents Based on Natural Language Processing.” Journal of the American Medical Informatics Association 11, no. 7 (2004): 392–402.
- “Testimony of the American Health Information Management Association to the Standards and Security Subcommittee of the National Committee on Vital and Health Statistics.” July 26, 2005. Available from the AHIMA Web site at http://www.ahima.org/dc/CAC_NCVHStestimony072605.asp (retrieved July 20, 2007).
- Friedman, Beth. “Coding Technology Today.” Journal of AHIMA 77, no. 4 (2006, April): 66–68.
- Ryan, G., and H. R. Bernard. “Techniques to Identify Themes.” Field Methods 5, no. 1 (2003, February): 85–109.
- Rogers, E. M. Diffusion of Innovations. New York: Free Press, 1995.
- Benson, S. “Computer-assisted Coding Software Improves Documentation, Coding, Compliance, and Revenue.” Perspectives in Health Information Management, Computer-assisted Coding Conference Proceedings, Fall 2006. Available from the AHIMA Web site at http://library.ahima.org/xpedio/groups/public/documents/ahima/bok1_032011.pdf (retrieved July 20, 2007).
- Servais, C. “Computer-assisted Coding for Inpatients: A Case Study.” Perspectives in Health Information Management, Computer-assisted Coding Conference Proceedings, Fall 2006. Available from the AHIMA Web site at http://library.ahima.org/xpedio/groups/public/documents/ahima/bok1_031998.html (retrieved July 20, 2007).
Article citation: Perspectives in Health Information Management, CAC Proceedings; Fall 2008