Automated Coding, Billing, and Documentation Support for Endoscopy Procedures

Home , Endoscopic ultrasound, Enteroscopy

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University

Kevin A. Jones

Graduate Program in Allied Medicine

The Ohio State University

2012

Master's Examination Committee:

Emily Patterson, PhD

Melanie Brodnik, PhD

Albert Lai, PhD

Kevin A. Jones

2012

Abstract

Clinical documentation has become extremely robust over the last decade with sophisticated algorithms to codify data for quality and operational improvements; however some electronic medical record systems are now adding billing data to create charge by documentation. This not only saves time for staff coding cases but creates an important link between the clinical documentation and charges. This one year retrospective case study analyzes one such application that codes endoscopy procedures based on the documentation to compare the automated coding method with manual physician billing to determine the efficacy of charge by documentation.

Dedication

Thank you to Craig who suffered through this process with me as well as the „friendly competition‟ with Matt and Maria to help me keep motivated.

iii

Acknowledgments

Thank you to Dr. Emily Patterson who really helped me through this entire process. I have found her extremely informative as a professor and advisor, but also professionally. Her enthusiasm combined with a deep knowledge of the healthcare environment is such an asset to this program.

I would also like to thank Dr. Melanie Brodnik who went out of her way to make this curriculum useful to someone with a different background such as mine. Her flexibility has challenged me with this endeavor which allowed me to study areas outside my comfort zone.

I would like to thank Dr. Albert Lai who I had the pleasure to work with on an academic and professional level before I started this program. He is so well-versed, accomplished and articulate in the biomedical informatics field, yet he is completely willing to lend his time and knowledge to you whenever asked.

And finally to Dr. Susan White who really put the meat behind the detail with her knowledge of both the statistics as well as the revenue cycle environment.

Vita

1993 B.S. Computer Science, Bowling Green State University

1993...... B.S. Computer Science, Bowling Green State University

1993 – 1996...... Programmer, The Ohio State University Medical Center,

Information Systems

1995 – 2002...... Consultant, ApotheTech

1996 – 2006...... Manager, The Ohio State University Medical Center,

Information Systems

2006 to present ...... Assistant Director, The Ohio State University Medical

Center, Information Technology

Publications

Mekhjian, H., Vasila, M., & Jones, K. (2008), Combine and conquer: computing from a

single database. Physician Executive, 34(5), 30.

Fields of Study

Major Field: Allied Medicine

Table of Contents

Abstract ...... ii

Dedication ...... iii

Acknowledgments ...... iv

Vita ...... v

List of Tables ...... x

List of Figures ...... xi

Chapter 1: Introduction ...... 1

Background and Setting ...... 1

Purpose of Study ...... 7

Significance of Study ...... 7

Conceptual Frame of Reference ...... 7

Research Questions ...... 8

Definition of Terms ...... 9

Limitations of the Study ...... 11

Chapter 2: Review of Literature ...... 12 vi

Current State ...... 12

Manual Charge Entry ...... 13

Professional Medical Record Coders ...... 13

Charge Capture Applications ...... 14

Computer-assisted Coding using Natural Language Processing...... 15

Charge by Documentation ...... 15

Benefits of Charge by Documentation ...... 16

Future Predictions ...... 17

Chapter 3: Methodology...... 18

Research Design ...... 18

Population and Sample Design ...... 18

Representation of Sample...... 19

Data Collection Procedures ...... 19

Data Collection Instrument ...... 20

Data Analysis ...... 21

Chapter 4: Results...... 24

Representativeness of Sample ...... 24

Profile of Population ...... 25

vii

Research Question 1 – What is the difference in accuracy between CPT codes

generated from charge by documentation as compared to manual generation? ...... 26

Research Question 2 – What is the difference in revenue between CPT codes generated

from charge by documentation as compared to manual generation? ...... 27

Procedures billed with lower reimbursement ...... 28

Undocumented Maneuvers ...... 31

Screening versus Diagnostic Discrepancies ...... 33

Modifier Discrepancies ...... 35

Other CPT Discrepancies ...... 37

Research Question 3 – Does charge by documentation functionality result in the

capture of additional or more specific ICD-9 codes than ones generated manually? ... 38

Summary ...... 40

Chapter 5: Discussion and Recommendations ...... 42

Summary of Findings ...... 42

Conclusions ...... 43

Implications of Study ...... 44

Recommendations ...... 44

References ...... 46

Appendix A – CPT Codes and Descriptions ...... 50

viii

Appendix B – CMS Price Fee Schedule ...... 53

Appendix C – Clinic Manual Billing Form...... 56

List of Tables

Table 1- Volume of procedures from the clinical system ...... 24

Table 2- Missing data...... 25

Table 3- Coding discrepancies ...... 26

Table 4- Lower reimbursement coding ...... 28

Table 5- Screening versus diagnostic coding ...... 29

Table 6- EGD with dilatation ...... 30

Table 7- Colonoscopy with injection ...... 30

Table 8- Undocumented maneuvers ...... 32

Table 9- Physician Coding Discrepancies ...... 33

Table 10- Physician Coding Discrepancies ...... 34

Table 11- CPT mismatches for entirely different procedures ...... 36

Table 12- Documented cases that were not billed ...... 37

Table 13- CPT mismatches for entirely different procedures ...... 38

Table 14- Increased ICD-9-CM capture by charge by documentation ...... 39

Table 15- ICD-9-CM codes not captured by charge by documentation ...... 40

List of Figures

Figure 1 - Clinical Documentation Example ...... 3

Figure 2 - Screening colonoscopy workflow ...... 5

Figure 3 - Conceptual Framework ...... 8

Figure 4 – Sample of an automated CPT prompt ...... 34

Figure 5 – Sample modifier prompt ...... 35

Chapter 1: Introduction

Background and Setting

Electronic medical records (EMRs) have great potential to improve record keeping and billing performance. Anesthesia has been an early adopter for using EMR documentation to support coding and billing in healthcare. For example, one study in anesthesia found that automated alerts of potential documentation errors decreased records that could never be billed and the median time to correct documentation errors, leading to an estimated increase in revenue of $400,000 per year. (1) Similarly, automated reminders were found to increase compliance with documentation of arterial catheterization in the perioperative setting, with an estimated increase in revenue of

$40,500 per year. (2) Finally, a system that automatically extracted information from an

EMR was found to decrease charge lag by ten days, but no clinically significant impact on revenue in anesthesia. (3)

To date, there has been limited research on directly automating real-time coding, billing, and documentation while providing care to patients. In one systematic review (4),

113 studies of automated coding and classification systems were assessed. Overall, the review found that there was much variety in terms of how automated tools were used to

1 support coding and classification, and that it is difficult to conduct a meta-analysis across studies or infer from individual studies how they might generalize to other settings. In particular, the review found that the complexity of the task had an impact on performance. A methodological concern with many of the reviewed studies was that manual coding was often compared to automated coding based upon manual coders who were hired only for research purposes and might not have been representative of how the work would have been conducted in an actual time-pressured situation.

A natural experiment occurred at the Ohio State University Wexner Medical

Center where a better direct comparison could be made between automated coding, billing, and documentation as compared to manual coding. One endoscopy clinic used an application that had the functionality to do automated billing. This clinic continued to use paper charge sheets for billing, selected by the physician using checkmarks on a paper form in Appendix C. Therefore, the automatically generated billing codes could be directly compared to the manually generated codes that were actually used for reimbursement purposes. The purpose of this study was to evaluate discrepancies in billing codes and estimate the revenue impact from these discrepancies.

The capability for the software that was used is presented generally in Figure 1.

Figure 1 displays a typical type of data entry technique that prompts the user to select a specific clinical procedure. In this case, the user has already selected colonoscopy as the procedure so a list of indications for performing a colonoscopy is displayed in a menu format. The user selects the most appropriate, and more importantly, the most accurate

2 indication, which in this case is screening for colorectal cancer with an average risk. The phrase “Screening for colorectal malignant neoplasm” is added to the documentation due to this selection. In this case, there are four distinct data elements captured in one step.

The physician sees the phrase “screen for Colorectal CA, Average Risk”, the report reads a more clinical phrase of “Screening for colorectal malignant neoplasm”, the patient discharge instructions reads a more user-friendly phrase of “cancer screening colonoscopy” and the database stores a discrete value representing both of these with references to both English phrases. Reporting at the very lowest level is now available for data mining.

ICD9-CM Screening 4 Average Risk Screen for colon cancer Surveillance Increased Risk Screen for colorectal cancer

Screen for rectal cancer Abdominal pain Family hist – distant relatvie Gastrointestinal bleeding

Polyps

Abnormal imaging

Figure 1 - Clinical Documentation Example

Relatively easy to use menu items can be created to capture basic and standardized codes such as:

 Current Procedural Terminology (CPT) Codes: A code set maintained by the

American Medical Association (AMA) to describe medical services, procedures,

medical equipment, supplies and implants.

 CPT Modifiers: A code set appended to the CPT codes if applicable to identify

descriptors such as reduced or additional services, laterality, etc. (5) This could alter

the reimbursement of a CPT code. (6)

 International Classification of Diseases (ICD-9-CM): The coding system

maintained by the Centers for Medicare and Medicaid (CMS) in the United States to

classify diseases, complications and comorbidities. (6) This code set is based off the

ICD-9 code set maintained by the World Health Organization

The assignment of billing codes is not a trivial task since coding errors directly impact reimbursement and can easily cause compliance issues. To illustrate this complexity, Figure 2 displays a billing code assignment for a simple screening colonoscopy.

G0105 Colorectal cancer screening; Billing Code Assignment for Simple Screening Colonoscopy colonoscopy on individual not meeting No criteria for high risk High Risk? No Yes G0121 Screening Colonoscopy Was the exam Was a polyp Diagnosis code Yes No Biopsied? No Colorectal cancer screening; Indication: Age completed? removed? more than V76.51? colonoscopy on individual meeting criteria for high risk

45378 Colonoscopy, flexible, proximal to splenic flexure; diagnostic, with or Yes without collection of specimen(s) by brushing or washing, with or without colon Yes decompression (separate procedure) 45380 Colonoscopy, flexible, proximal to splenic Yes flexure: with biopsy, single or multiple

45385 Colonoscopy, flexible, proximal to Hot Snare (Cautery) splenic flexure; with removal of tumor(s), polyp(s), or other lesion(s) by snare technique No 45384 Colonoscopy, flexible, proximal to Hot forceps (Cautery) splenic flexure; with removal of How was it tumor(s), polyp(s), or other lesion(s) by removed? hot biopsy forceps or bipolar cautery

45383 Colonoscopy, flexible, proximal to splenic flexure; with ablation of Not Documented? tumor(s)1 polyp(s). or other lesion(s) not amenable to removal by hot biopsy forceps, bipolar cautery or snare technique No 73 – Discontinued before anesth Anesthesia Discontinued Given? Yes 74 – Discontinued after anesth

Reduced or Discontinued?

Reduced 52 Modifier Needed

Figure 2 - Screening colonoscopy workflow

The consequences of inaccurate coding, especially for reimbursement purposes, can be significant whether the inaccuracy relates to reimbursement of services provided.

Billing a CPT code that has a lower reimbursement rate has an obvious impact. On the surface this may seem innocuous but coded data is used for more than just the bill. CPT and ICD-9-CM codes can be powerful metrics used for quality measures, operational improvement and research. Inaccurate coding jeopardizes these downstream efforts. (7)

(8)

The converse is selecting codes billed at a higher reimbursement rate than what was actually performed. This is often due to maneuvers that are performed but not documented or simply human error in selecting the wrong codes but could be viewed as a way to maximize revenue as well. If there is no automated linkage between clinical documentation and billing data, incorrect codes are often not caught until audits are performed. This could include chart reviews by internal hospital quality departments or the Joint Commission or it could be a payer such as the Center for Medicare & Medicaid

Services (CMS) comparing the rates of codes across many like hospitals broken down by physician. If the average of someone taking a biopsy is 80-85% for a given procedure but one physician bills biopsies over 95% of the time, then a review may be warranted. If documentation within the medical record does not support the clinical code(s) used for billing, the compliance of the healthcare provider can be called into question. Fraudulent billing is extremely serious and can result in huge fines or even revocation of the physician‟s license. (9)

To produce accurate coding, a fairly new methodology called Charge by

Documentation has been adopted to tie the clinical documentation with the charge codes.

While several healthcare application vendors support this in a variety of ways, a standardized name has not really established. This is sometimes labeled a “coding engine” for marketing reasons but this is not reflected in literature.

Purpose of Study

The purpose of this study was to determine the efficacy of charging by documentation automation by verifying the charge data generated to what the physician selects manually on a charge sheet. It analyzes both CPT and ICD-9-CM codes as well as billing modifiers from a case study using an application that codifies the data immediately and stores the information as discrete data from its creation.

Significance of Study

The three data elements (CPT codes, CPT Modifiers, ICD-9-CM codes) are the main driver of revenue. If these are wrong, revenue and/or compliance are affected.

Even if the billing codes are correct but the clinical documentation does not support the billing, then the data is essentially wrong. If the charge by documentation is accurate, an automatic tie between clinical documentation and billing information is established which will increase compliance and has some level of impact to the revenue.

Conceptual Frame of Reference

There are several ways that billing codes including CPTs, ICD-9-CM codes and modifiers are added to a patient‟s bill. The first two represent the traditional methods including manual charge sheets and having a certified medical record coder review the documentation and add codes to the case. In the last few years automation tools and highly defined content have been created to add billing detail while meeting CMS and

7 compliance standards. Figure 3 demonstrates the typical types of assignment of billing codes.

Conceptual Framework for Procedure Billing Clinical Documentation Approval Data Entry

Manual Charge Entry Physician completes a paper Clerk enters charges manually charge sheet

Medical Records Coder Certified Medical Records Coder reviews documentation Coder enters charges manually Physician creates a procedure and codes billing detail note including: Automation Process Charge Application  Exam being performed  Indications Interface pulls patient data and Physician has list of procedures  Impression clinical documentation into an to be billed and adds billing Interface to billing application  Maneuvers application detail  Medications/sedation  Findings Computer-assisted Coding  Recommendations Application “reads” Certified Medical Records  Signature documentation and parses strings Coder reviews documentation Interface to billing application to create codes and codes billing detail

Charge by Documentation

Physician reviews billing codes, Clinical Documentation content modifies if necessary and Interface to billing application generates billing codes for review approves

Figure 3 - Conceptual Framework

Research Questions

The goal of this study is to compare the revenue impacts of charge by documentation versus traditional physician paper billing. To do this, it will answer the following questions:

1. What is the difference in accuracy between CPT codes generated from charge by

documentation as compared to manual generation?

2. What is the difference in revenue between CPT codes generated from charge by

documentation as compared to manual generation?

3. Does charge by documentation functionality result in the capture of additional or

more specific ICD-9 codes than ones generated manually?

Definition of Terms

Colonoscopy – A medical diagnostic procedure that extends an endoscope through usually the anus to view the rectum, colon and large intestines for abnormalities.

Coding by Documentation – The function of an application to generate billing codes as a provider is creating clinical documentation.

Computer-assisted Coding – Employing an application to analyze the text of a report and generate billing codes based off of the identification of certain words or key phrases.

CMS – Centers for Medicare and Medicaid, which is the federal agency within the

United States Department of Health and Human Services (DHHS) responsible for overseeing the Medicare programs nationally as well as works with state-level Medicaid programs, HIPPA and other healthcare agencies.

CPT – Commonly used abbreviation for Current Procedural Terminology, which is a code set created by the American Medical Association which describes medical procedures and is used to bill insurance payers.

EGD – Esophagogastroduodenoscopy, which is a medical diagnostic procedure that extends a endoscopy through the mouth to the stomach and/or duodenum to view the upper digestive track for abnormalities. 9

EMR – Electronic Medical Record, also commonly called an Electronic Health Record

(EHR). This is a generally used term to describe applications that generate and/or store patient records.

Flexible Sigmoidoscopy – A medical diagnostic or screening procedure often called a

“flex sig” that extends an endoscopy through the anus to view the anus, rectum and prostate for abnormalities.

HIPAA – Health Insurance Portability and Accountability Act which was passed in 1996 in an effort to increase the access to healthcare in the United States.

ICD-9 – International Classification of Diseases, Ninth Revision, Clinical Modification is a diagnosis code set based on the ICD-9 codes created by the World Health Organization to describe diagnosis and procedure codes.

ICD-9-CM – International Classification of Diseases, Ninth Revision, Clinical

Modification is based upon the ICD-9 code set maintained by the Centers for Medicare and Medicaid (CMS) for use in the United States.

NCCI Edits – National Correct Coding Initiative effort started in 1996 by CMS to add rules into the diagnosis coding process to identify errors.

Revenue Cycle – The overarching collection of processes within a medical facility such as scheduling, registration, charge capture, billing that handles the financial aspects of patient care.

Limitations of the Study

Although many of the discrepancies are likely due to human error on the part of the physicians filling out the charge sheets or the clerks typing in the information, it is not possible to discern this based upon the study methods. Therefore, it is only possible to infer that there is a maximum error rate for either approach based upon the discrepancy rate.

This case study was performed at an ambulatory clinic, which provides low acuity exams. Patient conditions such as Barrett‟s, Crohns disease, etc. are generally not treated in this clinic as well as patients not able to undergo sedation. This study is limited to outpatients since inpatients are billed by Diagnosis Related Groups (DRG) as opposed to

CPT codes, thus making the comparison much more difficult. It is not expected that the outcome would change but could be an interesting follow-up study.

The study also limits itself to endoscopic procedures. This is meant to serve as a case study and could be expanded to additional procedure areas in the future.

Chapter 2: Review of Literature

Current State

Streamlining the revenue cycle is one of the top priorities at most medical facilities. As insurance payers are reducing payments and the government is pushing incentives for Meaningful Use with EMRs, the revenue cycle is often squeezed to optimize revenue while meeting ever-growing compliance regulations. (10) (11) To do this effectively, capturing the correct CPT and ICD-9-CM codes on a timely basis along with any appropriate modifiers to help ensure compliance is paramount. Insurance payers are requesting documentation to support the coding at a growing rate and variation can result in underpayment or a denial. Technology improvements on the payer side also means recognizing trends for medical facilities that may not be compliant so audits can be better targeted. To improve the chances of receiving a timely reimbursement means sending a clean claim.

Fortunately, technology for capturing clean CPT codes, CPT modifiers and ICD-

9-CM codes have also advanced allowing more options for medical facilities to implement. In addition to manual charge entry and medical record coders, a fairly broad range of application technology exists in the market to help alleviate the strain. Currently there are several methodologies behind associating billing codes such as manual charge

12 entry, medical record coders, charge capture applications, computer-assisted coding and charge by documentation.

Manual Charge Entry

Manual Charge Entry is the oldest and most straightforward method of selecting charge codes simply by checking options on a sheet of paper that includes all possible, or at least the most common billing codes used by a healthcare provider. Problems with this method of charge entry include the following: the physician does not complete the manual paper step, the wrong checkbox may be selected or unclear or the paper form may be misplaced before it is sent to a data entry clerk. In addition, these forms tend to be reproduced in bulk and left unmodified when billing codes are updated. It is, however, the least complicated and still widely used. Appendix C is the billing form that is used as a control for this case study.

Professional Medical Record Coders

There are a variety of certified medical record coding professionals that have the authority to review the clinical documentation and make a determination on the billing codes to be placed on a patient‟s bill. This represents a huge benefit over manual charge entry in the fact that these professionals are required to keep current with billing updates, requirements and compliance. In addition, they review the entire medical chart to review items like comorbidities, family history and previous surgeries. A coder would be able to

13 distinguish the difference between a high risk colonoscopy vs. an average risk which might go unnoticed by the physician. While coders are generally unable to make clinical decisions based off the documentation, they often identify potential errors and start a dialog with the provider for clarification when needed. The downside to including medical record coders is the obvious additional labor costs and a delay in the turnaround time to send a claim and receive payments.

Charge Capture Applications

Software applications have been developed to aid in charge capture including

PatientKeeper® (www.patientkeeper.com), Ingenious Med™ (www.ingeniousmed.com) and MedAptus© (www.medaptus.com). In the most simplistic way, it replaces the manual paper form with an application screen but is similar in nature. These applications however generally replace checkboxes with dropdown coded entries, on-line help, filtering and some level of business logic to guide physicians to selecting the correct entries. Applications such as these tend to have patient data interfaces and/or clinical documentation entries to prevent lost charges. A physician will login and find a list of cases that need to be coded. This dramatically improves the capture and accuracy of billing but does have some limitations. The physician needs to accurately select the correct billing codes based on what was performed. (12)

Computer-assisted Coding using Natural Language Processing

An ever-growing field of study called Natural Language Processing (NLP) includes building applications that in essence “reads” the clinical documentation to find enough segments of words necessary to generate CPT and diagnosis codes. (13) It does this by parsing common vernacular for a particular clinical specialty and segmenting sentences into “tokens” that are matched to codified elements. Computer-assisted coding applications such as LifeCode® by OPTUMInsight™ (www.OptumInsight.com),

Quantim® by QuadraMed© (www.quadramed.com) and 3M Coding and Reimbursement

Systems by 3M™ (www.3Mhis.com) use NLP to scan clinical documentation to generate

CPT and diagnosis codes. While the field has become very sophisticated over time and is in use commercially, the success rate is not very high and usually requires a medical record coder to review difficult cases and perform quality assurance on most or all coded documents and therefore mitigating the benefits. (14) (15) (16) (17) (18)

Charge by Documentation

Finally, the topic that is the focus of this study is charge by documentation. As opposed to the previous methods for charge capture, which rely on a separate step by the physician or a medical record coder, charge by documentation creates codes as the physician is documenting the case. This methodology is used by several procedure documentation systems such as Provation MD® by Wolters Kluwer® (www.Provation

Medical.com), EndoSoft® (www.endosoft.com) and gGastro® by GMed®

(www.gmed.com). In the application being reviewed, documenting is a series of menu- 15 driven choices, which equate to phrases concatenated to create the major sections such as the impression, findings, maneuvers, etc. After the clinical portions have been documented, CPT and diagnosis codes are presented based off these phrases. The physician has the option of modifying the pre-selected codes if necessary. They are also prompted if a billing modifier is appropriate. For instance, if a scope for a colonoscopy was only able to reach the rectum because of poor preparation by the patient, the physician would select the proper “extend scope to” field. The coding module would recognize this and prompt the user for a “52 – Reduced Services” modifier.

Benefits of Charge by Documentation

Charge by documentation holds significant improvements over the other billing mechanisms such as:

1. The billing codes by nature match the documentation, which ensures compliance.

If a physician overrides the codes for any reason, it is marked as such and can be

audited. Physicians can be trained that if the codes are not correct then they

should check the documentation.

2. Workflow is improved by the fact that billing and documentation are done as a

single fluid process. It is not uncommon for physicians to document and bill the

case before they leave the room, which improves accuracy.

3. The only real way to accidentally omit billing is to forget to document the case

entirely. This is very unlikely because the application is used intra-procedurally.

4. Assuming the software is maintained correctly, CPT, ICD-9-CM and billing

modifiers are automatically kept up-to-date.

5. The phrases are coded as dictionary items and need no parsing which eliminates

double meanings and variations in human language. The phrases can translate to

other languages as well as what is displayed on the document but the discrete

database elements are not language dependent making it transferrable to other

languages as well.

Future Predictions

EMR vendors see automatic charge capture and/or charge generation based off the provider‟s documentation as a huge selling point. Increasing this selling point means bigger profits and will significantly aid in the return on investment for the medical facility purchasing the software. In addition, the mandate to implement ICD-10-CM means all applications using ICD-9-CM need to rethink their current processes since the changes are so drastic between the two code sets. This gives a push to include charge capture as a part of this rewrite.

Chapter 3: Methodology

The purpose of this chapter is to present the methodology used to conduct this study. The chapter is divided into sections. The first section describes the research design of the study followed by a discussion of the population and sampling design in the second section. The third and fourth sections detail the data collection processed followed by a discussion of the data collection instrument and process for calculating revenue rates.

Research Design

This is a retrospective, cross sectional research design that analyzes endoscopy exams for a single location for one year that compares CPT, CPT modifier and ICD-9-

CM codes generated from a charge by documentation application with codes manually entered into an electronic billing system from paper charge sheets. This research study was approved by the Institutional Review Board at The Ohio State University.

Population and Sample Design

This case study uses one endoscopy clinic in a suburb of Columbus, OH, which performs approximately 3,000 exams a year. The clinic purchased a commonly used application in 2003 that includes charge by documentation functionality. Despite having 18 the ability to automate coding, they have not implemented this feature and have remained on paper charge sheets. The physician will manually checkmark all billing codes into the form displayed in Appendix C which is then entered into the professional practice billing system a day later by a data entry clerk. There is no medical record coder involved in this process. Meanwhile, the endoscopy application used to document the procedure note is creating billing codes as part of its standard functionality. Since the charge by documentation feature is standard in the application, it cannot be removed or turned off.

All CPT, ICD-9-CM and billing modifiers have been generated and even approved by the physician but never used by the clinic for billing purposes. More importantly, the workflow at this clinic prevents the codes from being used. This environment creates two distinct sets of billing data that are never cross-referenced but have the same data elements that can be easily compared.

Representation of Sample

The sample includes 100% of the patients seen in the ambulatory endoscopy clinic for fiscal one year, July 1, 2010 to June 30, 2011. Any cases that are missing data and cannot be compared will be clearly identified.

Data Collection Procedures

There were two extracts involved in this study:

1. Endoscopy Application Extract: This file includes the patient‟s medical

record number (MRN), date of service, diagnosis, procedure as well as the

CPT, ICD-9-CM codes and any related CPT modifiers automatically

generated from the system. The data in this file represents what would have

been coded by the system based off the clinical documentation if the clinic

would have implemented a workflow to use the data.

2. Professional Practice Billing Extract: The second extract is essentially the

same data set from the professional practice billing system. This represents

the data that was captured from the manual charge sheet displayed in

Appendix C and hand-keyed into the system by a data entry clerk.

Both data sets were downloaded to a single Excel spreadsheet. At no time was any patient identifiable information exported with the exception of the MRN, which was only used for data linking. The date of service was used to verify that the correct cases were linked since some patients have multiple endoscopic procedures in the same timeframe. By nature, it is not possible to have the same endoscopy exam performed multiple times on the same day without being part of the same procedure.

Data Collection Instrument

Microsoft Excel and the two data exports were the only instruments involved with this study. A database was purposely not created to limit potential for error in creation, importing and analysis. Instead, every record was compared for matches, discrepancies 20 and trends. Tabs within the spreadsheet contain the original unmodified data from both systems for reference. A working tab was added to combine the data in a logical order for analysis. Balancing functions have been added to the spreadsheet wherever possible to help ensure accuracy.

Data Analysis

MRN and dates of service from the two exports were linked to identify matches as well as discrepancies between the two code sets and incorrect coding. The two sets of codes (auto-generated and manually coded) were categorized into 4 groups:

1. Codes that match exactly. This indicates the auto-coding is equally effective

as manual coding.

2. Auto-generated codes that had a higher reimbursement rate than the manually

inputted codes.

3. Auto-generated codes that had a lower reimbursement rate than the manually

inputted codes.

4. Other discrepancies such as omission of charges, incomplete documentation,

etc.

This study assumes the clinical documentation is the gold standard. If a maneuver was performed but omitted in the documentation then it still cannot be billed. The fundamental principle of charge by documentation is to generate billing codes based on what was documented. While it should hold true that charge by documentation is the 21 gold standard since documentation drives the billing codes with this process, it will not be assumed that the auto-generated codes are correct. All discrepancies will require the documentation to be reviewed to guarantee that the physician documented correctly in order for the computer to generate the accurate codes. For instance, if a physician billed a biopsy but the computer did not auto-generate a CPT with a biopsy, then a review is needed to verify the biopsy was included with the documentation. The review will verify a correct code and identify the reasoning for any errors.

Payment Estimation Procedure

To determine if there was a difference in payments between the auto-generated and manual codes, estimates were generated based on publicly available national rates from Medicare. The measure for the payment variable was US dollars, rounded to the nearest dollar. In reality, the actual payment would depend upon local details, such as the insurance payer, locality and/or individual managed care contracts. Medicare national rates are arguably the most general since they are generated from the US government and often serve as the basis for private and other governmental payer rates. The rates were downloaded from the 2011 Physician Fee Schedule Search from the CMS website and are attached in Appendix B. (19) Finally, the estimate was based on the physician fee and not the facility fee, which generally does not vary between similar types of cases.

When multiple CPT codes are billed for the same date of service and are considered or significant procedures, which include all endoscopy services, the highest relative value unit (RVU) is billed at 100% and the other CPTs codes are billed with the

22 difference between the “endoscopy base” and the secondary CPT. For instance, if a colonoscopy is performed with a biopsy (45380) and a polypectomy (45385), which both have the “base endoscopy code” of a diagnostic colonoscopy (45378,) then the highest

RVU, in this case the 45385, would be billed at 100% but the 45380 would be billed using the physician fee schedule of $266.35 (45380) - $222.52 (45378) or an additional

$43.83. The total payment in this case would then be $316.03 + $43.83 or $359.86. (20)

(21) All payment calculations use this formula.

Chapter 4: Results

The purpose of this study was to determine the efficacy of charging by documentation automation by verifying the charge data generated to what the physician selects manually on a charge sheet. All matches and discrepancies between the auto- generated codes and manually inputted codes were categorized and analyzed. Any discrepancy was validated by reviewing the clinical documentation to determine if the auto-generated codes were appropriate and/or if the physician did not charge according to the documentation.

Representativeness of Sample

There were 2,921 total exams for the fiscal year between July 1, 2010 and June

30, 2011 for the endoscopy clinic as reported by a standard report in the endoscopy application and listed in Table 1.

Procedure Total Count Colonoscopies 1926/2,921 (65.9%) Upper GI Endoscopy 956/2,921 (32.7%) Flexible Sigmoidoscopy 30/2,921 (1.0%) Non-endoscopic G-Tube Exchange 5/2,921 (0.2%) Small Bowel Enteroscopy 4/2,921 (0.1%) Table 1- Volume of procedures from the clinical system 24

For balancing purposes, the same criteria generated 2,923 billed cases from the physician billing system. The discrepancy is reconciled by 2 cases that were billed but not documented at all in the endoscopy system.

Profile of Population

Out of the 2,923 cases studied, there were 51 cases in Table 2 that were missing data and removed from analysis making 2,872 the denominator for CPT calculations.

There is no reason to believe these 51 cases indicate bias with this study.

Procedure Total Rationale Count No documentation or 31 Since there is no supporting documentation, it is bill created impossible with the current data set to determine if the patient did not show, arrived but refused the test, arrived but was unable to undergo the test or another reason. It does not appear that the test was performed so it does not affect the outcome. Manual Edit 14 The endoscopy system marks a case if the physician overrode the auto-generated coding. Since the purpose of the study is to compare auto-generated coding with manual coding, overridden cases are outside the scope. Erroneous modifier – 2 PT is a new modifier added shortly after this study. PT Two cases used it early, which is appropriate and is not an error. Erroneous modifier – 1 Any case can have a modifier added and a single case LT had the LT modifier added to a colonoscopy. This is a likely data error System Issue 1 A single case was corrupted and not retrievable in the system. Billing credited 2 The billing office prevented sending of the claim for unknown reasons. Total 51 1.7% Table 2- Missing data 25

Research Question 1 – What is the difference in accuracy between CPT codes generated from charge by documentation as compared to manual generation?

A total of 350 discrepancies in CPT codes in Table 3 were identified from the total of 2,872, which is a 12.2% error rate. In a few cases, the discrepancies are minor and can be attributed to human error, but a majority of these cases either indicate a possible compliance issue such as billing for a CPT that is more expensive than the documentation can support or it is a loss of revenue because the CPT code was not accurate based off the documentation.

Discrepancy Total Rationale Count Coding with a lower 117 CPT codes billed were lower reimbursement from reimbursement rate what was documented indicating revenue loss Coding with a higher 102 CPT codes billed were higher reimbursement reimbursement rate from what was documented indicating compliance issues Coding discrepancy 45 The computer prompted the physician for a decision; the physician billed the different choice than what was answered in the computer Compliance 19 Irreconcilable differences between billed and discrepancies auto-generated CPT codes Missed billing 9 Complete revenue loss Billed cases, no 2 Cases that resulted in claims being sent but documentation documentation is not in the electronic system Modifier discrepancy 56 Discrepancy with billing modifiers that may result in revenue or compliance issues. Total 350 Table 3- Coding discrepancies

Research Question 2 – What is the difference in revenue between CPT codes generated from charge by documentation as compared to manual generation?

There was no statistically significant difference in payments based upon the automatically generated documentation as compared to what the physicians manually selected on a paper form. A 95% confidence interval for the average payment shift per claim is (-$0.71, $2.02). A hypothesis test showed that the average shift in payment per case is not significantly different from zero (p = 0.26).

Although there were no statistically significant differences, it is still instructive to explore the root cause of the discrepancies identified. A total of 117 of these cases resulted in a revenue loss of $2,145 because the CPT code billed was a lower reimbursement rate than the one that accurately matched the clinical documentation and

102 of these exams resulted in an inflated $4,669 because the codes were billed at a higher reimbursement rate than what the documentation supports. In addition, there were

9 cases that were documented but never billed, assumedly because the paper billing process was not completed by the time of this study‟s data collection, for a total of

$1,911. While all figures are professional fees only, another $3,168 in technical fees would have been lost due to the same manual process. Overall, there was a net loss in professional fees of $1,245 (95% confidence interval = -0.71 to 2.02. A hypothesis test showed that the average shift in payment is not significantly different from zero (p =

0.29). This however reflects the charges being billed and doesn‟t account for actual reimbursement which would factor denials and partial payments.

Procedures billed with lower reimbursement

Much like any charge by documentation application; the endoscopy system relies on specific selection items to be documented in order to select the correct CPT code. In the following examples, the physician documented polyp removals, stomas, biopsies or injections in the endoscopy system but did not bill for the corresponding CPT code on the manual charge sheet resulting in a lower reimbursement rate. Tables 4, 5 and 7 display how using the auto-generated code from the endoscopy system would have generated an additional $2,145.

Documentation Count Billed Est Correct Est Diff CPT Reimb CPT Reimb Foreign body removal 1 43235 $149 43247 $242 $93 Stoma not documented 1 44388 $163 45378 $221 $58 Stoma not documented 2 44389 $376 45380 $529 $153 Polyp removal by cold 7 45378 $1546 45380 $1853 $306 forceps Polyp removal by hot snare 4 45378 $890 45385 $1264 $374 Polyp removal by hot snare 3 45380 $1059 45385 $1256 $197 Distinct polyp removal by 1 45380 45380, hot snare and biopsy 45385-59 Polyp removal by hot snare 2 45384 $556 45385 $632 $76 Polyp removal by hot snare 1 45385 $316 45380, $516 $199 and cold forceps 45385 Total 22 $1456 Table 4- Lower reimbursement coding

In addition, if a patient was being seen for a screening biopsy but the physician performed a maneuver like a polyp removal then it should be billed at the higher rate as seen in Table 5. (22) In 76 cases, the physician documented a maneuver but continued to bill a screening code as opposed to the diagnostic CPT. In the case of G0105 and G0121, 28 the reimbursement didn‟t change because CMS rates are the same but it did result in a lower reimbursement for the other discrepancies.

Documentation Ct Billed Est Correct Est Diff CPT Reimb CPT Reimb Flex Sig screening 1 G0104 $62 45378 $221 $158 High-risk screening w/biopsy 10 G0105 $221 45378 $221 $0 Low-risk screening requiring 62 G0121 $4859 45378 $4859 $0 biopsy Low-risk screening requiring 3 G0121 $663 45380 $794 $131 polyp removal Total 76 $289 Table 5- Screening versus diagnostic coding

One common coding discrepancy revolves around the use of balloon dilatation.

The dilatation is in addition to the general EGD and not billed as a separate procedure.

Because of this, a 59-Distinct Service is not required unless the dilatation is done in a different area than the biopsy included in the 43239 CPT code. It is not possible to determine the intent of the physician based off the documentation and billing codes, nor is it possible to determine if the charge by documentation omitted a modifier. Estimated revenue loss or gain in Table 6 cannot be calculated in this case although it can be argued that the documentation does not match the billing codes and claims could be denied.

Documentation Count Billed Est CAC Est Diff CPT Reimb CPT Reimb EGD with dilatation 6 43249 43239, Unable to 43249 determine if the EGD with dilatation 1 43239 43239, physician billed 43249 incorrectly or did EGD, dilatation with 3 43239, 43239, not document so distinct services 43249- 43249 the computer 59 could determine. Total 8 Table 6- EGD with dilatation

Very similar to the balloon dilatation, physicians will often inject a segment of the colon with a permanent India ink „tattoo‟ in order to track locations for future procedures.

This procedures listed in Table 7 are in addition to the colonoscopy and should be billed as separate exams. The multiple endoscopy rule does get applied however regardless of a

59-Distinct Service modifier which is not required. (22)

Documentation Count Billed Est Correct Est Diff CPT Reimb CPT Reimb Colonoscopy with biopsy 1 45380 $265 45380, $295 $30 omitted injection 45381 Colonoscopy with biopsy 1 45380 $265 45380, $388 $123 omitted injection 45381, 45385 Colonoscopy with polyp 4 45385 $1256 45385, $1377 $121 removal omitted injection 45381 Colonoscopy with biopsy, 3 45380, $358 45380, $388 ($41) polyp removal by hot snare, 45385 45381, omitted injection 45385 Colonoscopy with biopsy, 1 45380, $357 45380, $388 $74 polyp removal by hot snare, 45385 45381, omitted injection 45385 Total 10 $307 Table 7- Colonoscopy with injection

Undocumented Maneuvers

In 102 cases or 3.5%, the clinical documentation does not support the CPT codes that were billed, thus the physician billed at a higher reimbursement rate than the documentation supported. This is a compliance issue that can result in corrective action.

It is not uncommon for insurance payers to request copies of the clinical documentation.

If this happens and the physician bills an EGD with a biopsy but the biopsy is not documented, it will often result in a complete denial of the case. This is not to say the discrepancy is intentional to increase revenue, but to identify cases where the documentation does not support the billing.

Table 8 calculates the difference between what the physician-billed versus the auto-generated CPT code from the application. This study does not include total or partial write-offs because the documentation does not match the billing codes. This should however be taken into consideration when evaluating a charge by documentation application and the total return on investment.

Documentation Ct Billed Est Correct Est Diff CPT Reimb CPT Reimb EGD Biopsy not documented 11 43239 $1921 43235 $1630 ($291) or removed with cold forceps 59-Distinct Services billed but 1 43239, $249 43251 $223 ($27) not documented 43251 Flex Sig Biopsy not 1 45331 $77 45330 $63 ($14) documented Foreign body removal not 1 45379 $276 45378 $221 ($55) documented Colon Biopsy not documented 3 45380 $795 45378 $663 ($132) Polyp taken with cold forceps, 3 45380 $795 45378 $663 ($132) not cautery Polyp noticed but not 1 45380 $265 45378 $221 ($44) removed; billed as if it were Colonoscopy through stoma 1 45380 $266 44389 $188 ($78) Polyp ablated with hot 2 45383 $682 45380 $529 ($153) forceps, not snare Polyp removal by hot snare 2 45383 $688 45385 $632 ($56) Polyp removed with hot 40 45384 $11036 45380 $10587 ($449) forceps, not snare 59-Distinct Services billed but 2 45380, $816 45384 $552 ($264) not documented 45384- 59 59-Distinct Services billed but 1 45380, $584 45385 $314 ($270) not documented 45384, 45385- 59 59-Distinct Services billed but 14 45380, $6247 45385 $4395 ($1852) not documented 45385- 59 Cold biopsy not documented 13 45380, $4651 45385 $4081 ($570) separate from lesion removal 45385- 59-Distinct Services billed but 2 45384, $892 45385 $628 ($264) not documented 45385- 59 Hot forceps not documented 3 45384, $1107 45385 $942 ($165) separate from hot snare 45385 Hot snare not documented 1 45384, $369 45385 $276 ($93) 45385 Total 102 ($4669) Table 8- Undocumented maneuvers

Screening versus Diagnostic Discrepancies

CMS has created unique codes for colonoscopy screening: G0105 for average risk and G0121 for high-risk. Any therapeutic maneuvers such as a biopsy or polyp removal can and should be billed with the standard diagnostic codes range of 45378-45385. If another finding is diagnosed, then the diagnostic code of 45378 should be used.

However, if no maneuvers are performed and there are no other findings, then the ICD-9 code of V76.51 should be used with the G0105 or G0121. (22) To accommodate this rule, the endoscopy application recognizes this scenario and prompts the physician to make a choice with supporting documentation. Figure 4 is a typical screen print of this functionality. In 26 cases listed in Table 9, the physician was prompted by the charge by documentation to make a decision based off the insurance company. In the system they chose one answer but marked the other on the billing sheet.

Documentation Count Gener Est Correct Est Diff ated Reimb CPT Reimb CPT Colonoscopy screening 1 45378 $221 G0105 $221 $0 billed as diagnostic Colonoscopy screening 24 45378 $5300 G0121 $5300 $0 billed as diagnostic Hot snare polyp removal 1 G0105 $221 45385 $314 $93 overridden with screening Total 26 $93 Table 9- Physician Coding Discrepancies

CPT code discrepancy

The documentation suggests that a diagnosic maneuver was performed. Screening procedures should use the “G” code but diagnostic procedures should use the standard CPT code.

45378 – Colonoscopy, flexible, proximal to splenic flexure; diagnostic, with or without collection of specimen(s) by brushing or washing, with or without colon decompressions (separate procedure)

G0121 – Colorectal cancer screening; colonoscopy on individual not meeting criteria for high risk

Next Cancel

Figure 4 – Sample of an automated CPT prompt

While the screen in Figure 4 is built to allow only one option, the system is not foolproof. In 18 cases listed in Table 10, the physician overrode the system in the end to bill both a screening and a diagnostic code. This violates the National Correct Coding

Initiative (NCCI) “Mutually Exclusive” rule because the codes are “inherent” according to CMS and cannot be used together. (22) In this example, revenue is not calculated since the claim would likely be denied by CMS rules.

Documentation Count Generated CPT Correct CPT Colonoscopy screening and 7 45378, G0105 45378 diagnostic as distinct exams Diagnostic and screening 14 45378, G0121 45378 billed together Total 18 Table 10- Physician Coding Discrepancies

Modifier Discrepancies

A charge by documentation application can be very helpful in identifying when billing modifiers are necessary or at least possible. Figure 5 displays a screen that is prompted when a physician documents that an exam could not be completed for some reason. The system does not go as far as to attach a modifier manually, but does prompt to help ensure proper billing. Figure 5 is what the physician is prompted by to select a modifier if the documentation supports one. If no mention of a procedure abort or reduced services is documented, then this screen is omitted.

Dialog Title

This procedure was documented as being incomplete or aborted. The system had defaulted a 53 modifier for the professional charge (discontinued procedure) and a 74 modifier (discontinued procedure after anesthesia/scope insertion) for the technical charge.

Please modify or accept.

45378 – Colonoscopy, flexible, proximal to splenic flexure; diagnostic, with or without collection of specimen(s) by brushing or washing, with or without colon decompressions (separate procedure)

Professional Modifier Technical Modifier

5522 –– RReedduucceedd SSeerrvviicceess 5522 –– RReedduucceedd SSeerrvviicceess

5533 –– DDiissccoonnttiinnuueedd PPrroocceedduurree 7733 –– DDiissccoonnttiinnuueedd pprriioorr ttoo aanneesstthheessiiaa//ssccooppee iinnsseerrttiioonn

7744 –– DDiissccoonnttiinnuueedd aafftteerr ttoo aanneesstthheessiiaa//ssccooppee iinnsseerrttiioonn

Next Cancel

Figure 5 – Sample modifier prompt

In 41 cases listed in Table 11, Modifier 52 – Reduced Services or Modifier 53 –

Discontinued Services were omitted when it was documented in the procedure note. This has an effect on both revenue and compliance. The patient was billed for the full charge but only received some portion of the services. The 52 modifier should have been included which could have reduced the reimbursement by 50% although some insurance companies continue to reimburse for the full amount regardless. This error is often highlighted to insurance companies and auditors if the patient has a repeat exam weeks or months later because the first was aborted due to poor preparation. Compliance issues are very clear in this case and corrective action may be taken.

In 8 cases, Modifier 59 – Distinct Services was omitted. This happens when two similar cases are performed during one visit but they are done as a separate procedure.

The first case is reimbursed at 100% and subsequent cases are reimbursed at 50%. The reduction is due to some of the cost being shared among all procedures such as a single room utilization, sedation session, staff time, etc., but accommodates for extra work performed.

Modifier Description Revenue Impact Count 22 Documentation does not support No impact 2 52 Reduced services Some payers reduce payment by 24 53 Discontinued services up to 50%. Some do not. 17 Most payers will pay in full for two 59 Distinct services distinct procedures 8 Hospital discontinue before or after anesthesia. These should have been 73/74 coded with 52/53 for physician fees No impact 6 Total 56 Table 11- CPT mismatches for entirely different procedures

Other CPT Discrepancies

Paper billing sheets are prone to various unavoidable issues. Lost billing sheets accounted for 9 exams listed in Table 12 being completely missed. Since there was no bill generated at all, a facility fee will be estimated to account for the complete loss of revenue. A facility like this clinic will have a Medicare rate of $344 for EGDs and $362 for colonoscopies.

Billed Description Count Physician Facility Total Lost CPT Charge Fee Revenue 43235 Diagnostic EGD 1 $148 $344 $492 43239 EGD with biopsy 4 $699 $1376 $2075 45378 Diagnostic colonoscopy 1 $221 $362 $583 45380 Colonoscopy with biopsy 2 $529 $724 $1253 Colonoscopy with 45385 polypectomy by hot snare 1 $314 $362 $676 Total 9 $1911 $3168 $5079 Table 12- Documented cases that were not billed

There were 3 exams that were billed by the physician but there is no documentation to justify that the exam was performed. Since the application is the same creating the charges as well as capturing images, vitals, etc., it would be difficult for a case to be performed without any indication in the clinical system. Any request for clinical documentation would most likely result in the denial of the claim. In addition, there are 16 cases in Table 13 that are so mismatched that the CPT codes billed indicate a completely different exam than the ones documented.

Billed Description Documented Description Count CPT CPT 43200 Esophagoscopy 43235 EGD 3 43202 Esophagoscopy 43239 EGD 1 43235 EGD 43200 Esophagoscopy 1 43239 EGD 45450 Invalid code 1 43239 EGD 45380 Colonoscopy 2 44360 Small intestinal endoscopy 45378 Colonoscopy 1 45330 Flexible Sigmoidoscopy 45378 Colonoscopy 1 45331 Flexible Sigmoidoscopy 45380 Colonoscopy 2 45331 Flexible Sigmoidoscopy 45378 Colonoscopy 1 45378 Colonoscopy 43235 EGD 2 G0121 Colonoscopy 43239 EGD 1 Total 16 Table 13- CPT mismatches for entirely different procedures

Research Question 3 – Does charge by documentation functionality result in the capture of additional or more specific ICD-9 codes than ones generated manually?

Additional codes were generated from the charge by documentation as compared to the manual generation. It is inconclusive if charge by documentation creates more specific ICD-9-CM codes than manual paper billing.

Diagnosis code capture has a somewhat different meaning in the clinical system than it does in the billing system. While it is a required data element on any charge, a code of esophageal reflux will be reimbursed the same as esophageal reflux plus a hiatal hernia therefore reducing the significance of accuracy. However, it is clinically significant to document the hiatal hernia for patient care.

Physicians have a greater incentive to capture diagnosis codes in clinical documentation because of the patient care concerns. Out of the 2,923, the list of ICD-9-

CM codes between the endoscopy system and the paper charge sheet matched in 905 cases for an accuracy rate of 30.9%. The application generated more codes based off the documentation in 1,818 or 62.2% of the cases. Another 693 or 23.7% of cases resulted in the physician adding an ICD-9-CM code without adequate documentation. In the remainder of cases, at least one additional code and up to 8 addition codes were generated. Table 14 details the amount of additional ICD-9-CM codes that were auto- generated by the computer application.

Accuracy Metric Count Complete match between physician and auto-generated codes 905/2,923 (30.9%) Endoscopy application generated 1 additional ICD-9 code 702/2,923 (24.0%) Endoscopy application generated 2 additional ICD-9 codes 595/2,923 (20.3%) Endoscopy application generated 3 additional ICD-9 codes 310/2,923 (10.6%) Endoscopy application generated 4 additional ICD-9 codes 143/2,923 (4.8%) Endoscopy application generated 5 additional ICD-9 codes 47/2,923 (1.6%) Endoscopy application generated 6 additional ICD-9 codes 14/2,923 (0.5%) Endoscopy application generated 7 additional ICD-9 codes 4/2,923 (0.1%) Endoscopy application generated 8 additional ICD-9 codes 3/2,923 (0.1%) Total 2,723/2,923 (93.1%) Table 14- Increased ICD-9-CM capture by charge by documentation

A much lower percentage of cases in Table 15 had the physician billing for ICD-

9-CM codes that were not documented in the system. This means the physician billed using an ICD-9-CM code but there is no mention of this in the documentation.

Accuracy Metric Count Physician billed 1 ICD-9 code but did not document 598/2,923 (20.5%) Physician billed 2 ICD-9 codes but did not document 76/2,923 (2.6%) Physician billed 3 ICD-9 codes but did not document 18/2,923 (0.6%) Physician billed 4 ICD-9 codes but did not document 1/2,923 (0.0%) Total 693/2,923 (23.7%) Table 15- ICD-9-CM codes not captured by charge by documentation

Generally billing data is not used for clinical research however it should be noted that most billing systems only store up to four ICD-9-CM codes because that is the limit that can be billed on the Physician 1500 Bill or Universal Bill (UB). The clinical system does not have this limitation so 193 cases matched for the first 4 codes but the clinical system stored more than four. This was considered a reasonable limitation and not a discrepancy between charge by documentation and physician billing.

One observation that does not support the hypothesis is the matter of specificity of

ICD-9-CM codes. In 96 cases, the codes basically matched but were not specific enough.

For instance, the computer may have coded 786.5 – Chest pain where the physician billed

786.59 – Discomfort, pressure or tightness in chest. Neither is incorrect, one is just more specific. Out of the 96 discrepancies, the physician billed 48 of them using a more specific code and the computer billed 48 of them using a more specific code.

Summary

In summary, the findings were that there was a 12.2% discrepancy rate between the charge by documentation and manual generation of CPT codes, no statistically

40 significant difference on estimated revenue generation, no difference in specificity of

ICD-9 codes, and additional codes generated by charge by documentation.

Overall, the study suggests relying on the system for charge by documentation purposes, as long as quality improvements methods are employed by coding professionals to audit the process. Using the system will likely reduce time to reimbursement, fewer denials or reduced payments due to higher agreement between documentation and CPT codes, and will reduce time spent by physicians filling out charge sheets and clerks typing in data from the sheets.

Chapter 5: Discussion and Recommendations

Summary of Findings

This study was to determine the efficacy of charge by documentation when compared to manual billing by validating the charge codes against clinical documentation. A 12.2% discrepancy rate was found between the manually generated

CPT codes and billing modifiers as compared to charge by documentation. There was no practically or statistically significant effect on reimbursement with an annual difference of only $1,245 (95% confidence interval = -0.71 to 2.02, p = 0.29). This is misleading however since matching clinical documentation with charges should in theory reduce denials as well as reduce the labor necessary to manually code and bill charges. In addition, timeliness of charges can benefit the financial status of a practice.

A similar finding was realized with ICD-9-CM codes. The discrepancy rate between mutually generated codes and the charge by documentation was 66.1%. The direct linkages between the clinical diagnoses which drive the billing ICD-9-CM codes greatly improve the capture of data.

One observation is the fact that there was not a single billing code that the application generated that did not match the clinical documentation. All discrepancies were due to the fact that the manual charge sheet did not match the auto-generated codes,

42 not that the auto-generated codes did not match the clinical documentation. There are cases when the physician did not document a key aspect of a procedure that affected the coding, but the system was able to code correctly with the details that were given.

Regardless of the financial ramifications, this fact alone supports the conclusion that charge by documentation is effective.

The total return on investment would be a valid follow-up study. The financial data in this study is based solely on charges and does not take into consideration denials, write-offs or partial payments in relation to the lack or incompleteness of clinical documentation. In addition, the cost of such an application, both in initial acquisition costs as well as annual support costs needs to be factored into a total cost of ownership.

However, if the costs do not justify the application installation, then additional quality assurance checks need to be implemented to add a level of compliance.

Conclusions

Charge by documentation is effective and should be considered where economically feasible. Implementing a system to tie clinical documentation with billing data has a positive revenue and significant compliance impact. Automating this with a standard billing interface from the charge by documentation to the billing application reduces the reliance on physicians to do coding, a task for which they usually are not trained to do, and adds an audit trail to ensure compliance.

Implications of Study

As ICD-9-CM is nearing its end of life, ICD-10-CM expands the number of codes from 14,000 codes to 68,000, which increases the amount of complexity and training requirements. (23) Having an application that can accurately translate “Barrett‟s esophagus” (ICD-9-CM 530.85) to the new ICD-10-CM codes including Barrett‟s esophagus (K22.7), without dysplasia (K22.70), with low grade dysplasia (K22.710), high grade dysplasia (K22.711) or unspecified dysplasia (K22.719) based off of English phrases to reduce the implementation, maintenance and coding errors due to the ICD-10-

CM conversion is a valid approach.

Recommendations

The following recommendations have evolved from this study and are directed toward the importance of accurate documentation and coding for healthcare reimbursement and compliance purposes.

1. Whether it is a large-scale EMR application or a niche documentation system,

it is important to review the billing functionalities of the system. This can

often add a cost-savings factor as well as reduce denials by adding compliance

measures.

2. Avoiding antiquated manual paper billing processes are essential for

compliance measures and automated charge capture help in this goal.

Creating a link between the billing codes and documentation which is the gold

standard, is highly beneficial to healthcare organizations.

3. As with any clinical system, periodic quality assurance testing needs to be part

of the standard maintenance of the system in order to ensure accuracy of

documentation and adherence to coding compliance issues. Provider

education on accurate documentation, system prompts and billing rules help

ensure compliance and risks of denials, audits and reduced reimbursement.

References

1. Automated Documentation Error Detection and Notification Improves Anesthesia

Billing Performance. Spring, Stephen F, et al., et al. 1, s.l. : Anensthesiology, 2007,

Vol. 106.

2. Electronic Reminders Improve Procedure Documentation Compliance and

Professional Fee Reimbursement. Kheterpal, Sachin, et al., et al. 3, s.l. : Anesthesia &

Analgesia, 2007, Vol. 104.

3. Development of a Module for Point-of-care Charge Capture and Submission Using an

Anesthesia Information Manaagement System. Reich, David L, et al., et al. 1, s.l. :

Anesthesiology, 2006, Vol. 105.

4. A Systematic Literature Review of Automated Clinical Coding and Classification

Systems. Stanfill, Mary H, et al., et al. s.l. : JAMIA, 2010, Vols. 17:646-651.

5. U.S. Department of Health & Human Services. Overview. Centers for Medicare &

Medicaid Services. [Online] 8 1, 2011. [Cited: 10 2, 2011.] https://www.cms.gov/MedHCPCSGenInfo/.

6. World Health Organization. Classifications. World Health Organization. [Online]

2011. [Cited: 10 2, 2011.] http://www.who.int/topics/classification/en/.

7. Improving the Quality of the Coding of Primary Diagnosis in Standardized Discharge

Summaries. Avillach, Paul, Joubert, Michel and Fieschi, Marius. s.l. : Health Care

Manage Sci, 2008, Vol. 11.

8. Get Paid for What You Do: Dictation Patterns and Impact on Billing Accuracy.

Thorwarth Jr, William T. 2:665-669, s.l. : Journal of American College of Radiology,

2005.

9. Clinical Documentation, Coding and Billing. Swigert, Nancy B. 2, s.l. : Seminars in

Speech and Language, 2006, Vol. 27.

10. Computer Assisted Coding: The secret weapon. Morsch, Mark. 2, San Diego :

Health Management Technology, 2010, Vol. 31. 10744770.

11. Center for Medicare & Medicaid Services (CMS). CMS EMR Meaningful Use

Overview. EHR Incentive Programs. [Online] 4 20, 2011. [Cited: 8 14, 2011.] https://www.cms.gov/EHRIncentivePrograms/30_Meaningful_Use.asp.

12. Implementing a computerized charge capture system to improve billing work flow and reduce errors in data entry. Jao, Chiang S, Helgason, Cathy M and Zych, Donna

A. 2, s.l. : Journal of Computers, 2009, Vol. 4.

13. Building a biomedical tokenizer using the token lattice design pattern and the adapted Viterbi algorithm. Barrett, Neil and Weber-Jahnke, Jens. 51, s.l. : BMC

Bioinformatics, 2010, Vol. 12 (suppl 3).

14. Automatic construction of rule-based ICD-9-CM coding systems. Farkas, Richard and Szarvas, Gyorgy. s.l. : BMC Bioinformatics, 2008, Vol. 9 (Suppl 5).

15. Automation of a problem list using natural language processing. Meystre, Stephanie and Haug, Peter J. 30, s.l. : BMC Medical Informatics and Decision Making, 2005, Vol.

16. Application of Natural Language Processing to VA Electronic Health Records to

Identify Phenotypic Characteristics for Clinical and Research Purposes. Gundlapalli,

Adi V, et al., et al. 40, s.l. : AMIA, 2008, Vol. 36.

17. Using Intrinsic and Extrinsic Metrics to Evaluate Accuracy and Facilitation in

Computer-assisted Coding. Resnik, Philip, et al., et al. s.l. : Prospectives in Health

Information Management, CAC Proceedings, 2006.

18. A systematic literature review of automated clinical coding and classifcations systems. Stanfill, Mary H, et al., et al. s.l. : JAMIA, 2010, Vol. 17.

19. US Department of Health and Human Services. Physician Fee Schedule Search.

Centers for Medicare and Medicaid Services. [Online] January 18, 2012. [Cited: January

29, 2012.] http://www.cms.gov/apps/physician-fee-schedule/search.

20. Centers for Medicare and Medicaid Services. www.cms.gov. www.cms.gov.

[Online] January 13, 2012. [Cited: January 29, 2012.] http://www.cms.gov/manuals/downloads/clm104c04.pdf.

21. Bronchoscopy Billing: The Multiple Endoscopy Rule. Plummer, Alan M. s.l. :

Americal Thoracic Society Coding and Billing Quarterly, 2008, Vol. September.

22. Ingenix. Coders' Desk Reference for procedures 2011. s.l. : Ingenix, 2011.

1601514123.

23. American Medical Association. Transation Code Set Standards. American Medical

Association. [Online] [Cited: January 31, 2012.] http://www.ama- assn.org/ama/pub/physician-resources/solutions-managing-your-practice/coding-billing- insurance/hipaahealth-insurance-portability-accountability-act/transation-code-set- standards/icd10-code-set.page.

24. Julius - A template based supplementary electronic health record system. Chen,

Rong, Enberg, Gosta and OKlein, Gunnar. 10, s.l. : BMC Medical Informatics and

Decision Making, 2007, Vol. 7.

Appendix A – CPT Codes and Descriptions

List of Endoscopy CPT codes with exact descriptions from the American Medical Association:

CPT Description 0528F Recommended follow-up interval for repeat colonoscopy of at least 10 years documented in colonoscopy report (End/Polyp) 0529F Interval of 3 or more years since patients last colonoscopy, documented (End/Polyp) 3018F Pre-procedure risk assessment AND depth of insertion AND quality of the bowel prep AND complete description of polyp(s) found, including location of each polyp, size, number and gross morphology AND recommendations for follow-up in final colonoscopy report documented (End/Polyp) 44388 Colonoscopy through stoma; diagnostic, with or without collection or specimen(s) by brushing or washing (separate procedure) 44389 Colonoscopy through stoma; with biopsy, single or multiple 44390 Colonoscopy through stoma; with removal of foreign body 44391 Colonoscopy through stoma; with control of bleeding (eg, injection, bipolar cautery, unipolar cautery, laser, heater probe, stapler, plasma coagulator) 44392 Colonoscopy through stoma; with removal of tumor(s), polyp(s), or other lesion(s) by hot biopsy forceps or bipolar cautery 44393 Colonoscopy through stoma: with ablation of tumor(s)1 polyp(s), or other lesion(s) not amenable to removal by hot biopsy forceps, bipolar cautery or snare technique 44394 Colonoscopy through stoma: with removal of tumor(s), polyp(s), or other lesion(s) by snare technique 44397 Colonoscopy through stoma: with transendoscopic stent placement (includes predilation) 45355 Colonoscopy. rigid or flexible, transabdominal via colotomy, single or multiple 45378 Colonoscopy. flexible, proximal to splenic flexure; diagnostic, with or without collection of specimen(s) by brushing or washing, with or without colon decompression (separate procedure) 45379 Colonoscopy, flexible, proximal to splenic flexure; with removal of foreign body 45380 Colonoscopy, flexible, proximal to splenic flexure: with biopsy, single or multiple 45381 Colonoscopy, flexible, proximal to splenic flexure; with directed submucosal injection(s), any substance

45382 Colonoscopy, flexible, proximal to splenic flexure; with control of bleeding (eg. injection, bipolar cautery, unipolar cautery, laser, heater probe, stapler, plasma coagulator) 45383 Colonoscopy, flexible, proximal to splenic flexure; with ablation of tumor(s)1 polyp(s). or other lesion(s) not amenable to removal by hot biopsy forceps, bipolar cautery or snare technique 45384 Colonoscopy, flexible, proximal to splenic flexure; with removal of tumor(s)1 polyp(s). or other lesion(s) by hot biopsy forceps or bipolar cautery 45385 Colonoscopy. flexible, proximal to splenic flexure; with removal of tumor(s), polyp(s). or other lesion(s) by snare technique 45386 Colonoscopy, flexible, proximal to splenic flexure; with dilation by balloon, 1 or more strictures 45387 Colonoscopy. flexible, proximal to splenic flexure: with transendoscopic stent placement (includes predilation) 45391 Colonoscopy, flexible, proximal to splenic flexure; with endoscopic ultrasound examination 45392 Colonoscopy, flexible, proximal to splenic flexure; with transendoscopic ultrasound guided intramural or transmural fine needle aspiration/biopsy(s) G0105 Colorectal cancer screening; colonoscopy on individual at high Ask G0120 Colorectal cancer screening; alternative to G0105, screening colonoscopy, barium enema. G0121 Colorectal cancer screening. colonoscopy on individual not meeting criteria for high risk

Appendix B – CMS Price Fee Schedule

List of CMS National rates. See the methodology chapter for details on how this fee schedule is used to determine rates for multiple endoscopy services performed on the same date of service.

FACILITY ENDO Base Add CPT CPT SHORT DESCRIPTION PRICE BASE Fee payments 43201 Esoph scope w/submucous inj $131.49 43200 $106.35 $25.14 43202 Esophagus endoscopy biopsy $117.22 43200 $106.35 $10.87 43204 Esoph scope w/sclerosis inj $226.62 43200 $106.35 $120.28 43205 Esophagus endoscopy/ligation $228.66 43200 $106.35 $122.32 43215 Esophagus endoscopy $157.31 43200 $106.35 $50.96 43216 Esophagus endoscopy/lesion $146.10 43200 $106.35 $39.75 43217 Esophagus endoscopy $174.30 43200 $106.35 $67.95 43219 Esophagus endoscopy $174.98 43200 $106.35 $68.63 43220 Esoph endoscopy dilation $129.45 43200 $106.35 $23.10 43226 Esoph endoscopy dilation $144.40 43200 $106.35 $38.05 43227 Esoph endoscopy repair $214.39 43200 $106.35 $108.04 43228 Esoph endoscopy ablation $227.64 43200 $106.35 $121.30 43231 Esoph endoscopy w/us exam $194.35 43235 $148.14 $46.21 43232 Esoph endoscopy w/us fn bx $267.73 43235 $148.14 $119.60 43234 Upper gi endoscopy exam $122.99 43235 Uppr gi endoscopy diagnosis $148.14 43235 Base 43236 Uppr gi scope w/submuc inj $179.06 43235 $148.14 $30.92 43237 Endoscopic us exam esoph $240.55 43235 $148.14 $92.42 43238 Uppr gi endoscopy w/us fn bx $300.01 43235 $148.14 $151.87 43239 Upper gi endoscopy biopsy $174.64 43235 $148.14 $26.50 43240 Esoph endoscope w/drain cyst $405.00 43235 $148.14 $256.86 43241 Upper GI endoscopy with tube $158.67 43235 $148.14 $10.53 43242 Uppr gi endoscopy w/us fn bx $432.52 43235 $148.14 $284.38 43243 Upper gi endoscopy & inject $273.17 43235 $148.14 $125.03 43244 Upper GI endoscopy/ligation $301.71 43235 $148.14 $153.57 43245 Uppr gi scope dilate strictr $191.97 43235 $148.14 $43.83 43246 Place gastrostomy tube $256.86 43235 $148.14 $108.72 43247 Operative upper GI endoscopy $204.54 43235 $148.14 $56.40 43248 Uppr gi endoscopy/guide wire $192.31 43235 $148.14 $44.17 43249 Esoph endoscopy dilation $177.36 43235 $148.14 $29.22 54

43250 Upper GI endoscopy/tumor $192.65 43235 $148.14 $44.51 43251 Operative upper GI endoscopy $222.55 43235 $148.14 $74.41 44388 Colonoscopy $168.52 44388 Base 44389 Colonoscopy with biopsy $188.23 44388 45330 Diagnostic sigmoidoscopy $63.20 45330 Base 45331 Sigmoidoscopy and biopsy $76.79 45330 $63.20 $13.59 45332 Sigmoidoscopy w/fb removal $112.12 45330 $63.20 $48.93 45333 Sigmoidoscopy & polypectomy $111.78 45330 $63.20 $48.59 45334 Sigmoidoscopy for bleeding $166.82 45330 $63.20 $103.63 45335 Sigmoidoscopy w/submuc inj $93.44 45330 $63.20 $30.24 45337 Sigmoidoscopy & decompress $144.40 45330 $63.20 $81.20 45338 Sigmoidoscopy w/tumr remove $143.72 45330 $63.20 $80.52 45339 Sigmoidoscopy w/ablate tumr $189.93 45330 $63.20 $126.73 45340 Sig w/balloon dilation $117.56 45330 $63.20 $54.36 45341 Sigmoidoscopy w/ultrasound $160.03 45342 Sigmoidoscopy w/us guide bx $244.29 45345 Sigmoidoscopy w/stent $177.70 45330 $63.20 $114.50 45355 Surgical colonoscopy $206.58 45378 Diagnostic colonoscopy $220.85 45378 Base 45378 Diagnostic colonoscopy $63.20 45378 Base 45379 Colonoscopy w/fb removal $276.23 45378 $220.85 $55.38 45380 Colonoscopy and biopsy $264.68 45378 $220.85 $43.83 45381 Colonoscopy submucous inj $251.09 45378 $220.85 $30.24 45382 Colonoscopy/control bleeding $337.73 45378 $220.85 $116.88 45383 Lesion removal colonoscopy $341.46 45378 $220.85 $120.62 45384 Lesion remove colonoscopy $275.89 45378 $220.85 $55.04 45385 Lesion removal colonoscopy $313.94 45378 $220.85 $93.10 45386 Colonoscopy dilate stricture $270.79 45378 $220.85 $49.95 45387 Colonoscopy w/stent $351.66 45378 $220.85 $130.81 G0104 CA screen;flexi sigmoidscope $63.20 G0105 Colorectal scrn; hi risk ind $220.85 G0121 Colon ca scrn not hi rsk ind $220.85

Appendix C – Clinic Manual Billing Form

Manual billing form used currently at the endoscopy clinic used in this case study.