What is ambulatory payment classification
Under inpatient prospective payment, only ICD-9-CM diagnosis and procedure codes are required to determine the Diagnosis Related Group (DRG), the basis for payment. The ICD-9-CM codes for DRG determination come from one source - coders in the Health Information Management Department. The Outpatient Prospective Payment System requires CPT codes for all services and supplies to determine the correct Ambulatory Payment Categories (APCs) to determine payment. The CPT-4 codes for APC determination come from at least two sources - coders in the Health Information Management Department and a Charge Master. In other situations, codes may also come from registration staff (who enter codes to determine the medical necessity of tests), from order entry systems, or from ancillary systems (e.g. laboratory, radiology, etc). The challenge is how to merge CPT codes from the various sources into one file so that the APC grouper software can analyze the data to determine the APCs to be assigned and the reimbursement to be expected.
There are several alternatives to address the issue of merging codes from various sources:
(1) Send all CPT codes to an abstract-based encoder/APC grouper
(2) Send all CPT codes to a billing-based grouper
(3) Send all CPT codes to a clearinghouse grouper
In this scenario, CPT codes from the charge master and/or ancillary systems are interfaced to a data file that can be accessed by HIM coders and are also interfaced to the encoder/grouper software based in the HIM system. The coders review the CPT codes entered by the charge master and other systems, add any CPT codes needed, and add modifiers, if necessary. The HIM coders also enter the ICD-9-CM diagnosis codes. Generally coders do not delete or change CPT codes generated by the charge master or ancillary systems, but bring questionable codes to the attention of patient accounts and/or the ancillary department responsible.
The advantages of this alternative are:
(1) HIM coders can review all codes and compare them with the medical record documentation. They can catch any services that may be mis-coded, services provided but not coded, and services coded that were not documented.
(2) HIM coders can add modifiers related to other codes. This would include modifiers for repeated procedures, modifiers for services performed in conjunction with clinic or emergency department visits, etc.
(3) Through notification to patient accounts or ancillary departments, HIM coders can effect correction of systemic coding problems.
The disadvantages of this alternative are:
(1) Coding must wait for charges to be posted. If the charges are not posted, CPT codes are not generated for review.
(2) Late charges may cause incorrect modifiers or a lack of modifiers to be assigned because HIM coders will not see the related CPT codes.
(3) If HIM notices about coding discrepancies are not investigated, the same coding problems can occur repeatedly, causing a reduction in coder productivity
(4) Extensive programming may be required to interface charge master and/or ancillary systems to the HIM-based encoder.
In this scenario, the codes from the HIM coders and the codes from the charge master or other sources are gathered in the bill processing software module. The APC grouper, interfaced with the billing database, would produce APC assignments as a by-product of the bill generation process.
The advantages of this alternative are:
(1) The data flow remains the same as that prior to APC implementation. Little or no rerouting of data is needed.
(2) HIM coding is not held up by late charges
The disadvantages of this alternative are:
(1) Unless someone reviews all the codes prior to bill transmission, there will be no review of the accuracy of the CPT codes as compared with the medical record
(2) Unless there is a pre-bill review, modifiers may be missed.
Some facilities prefer to send all billing data, including the CPT codes from all sources, to a clearinghouse. The clearinghouse will merge the codes and determine the APC for the provider as one of the services they provide.
The advantages of this choice are:
(1) No programming or changes in data flow are required
(2) The clearing house may include sophisticated edits on all codes/modifiers
The disadvantages of this choice are:
(1) Without sophisticated edits, the clearinghouse may miss modifiers
(2) No one in-house is able to review the accuracy of codes/APCs prior to billing data transmission
(3) If the clearinghouse sends error or edit reports back to the facility, the HIM staff may need to review the medical record twice to correct the code edits. This will decrease coder productivity and delay bill processing.
No one scenario presents a perfect solution. Providers must evaluate their current systems and interfaces to determine which is most cost effective.
A second challenge occurs because of the paper flow associated with outpatient services. Inpatient records are gathered from nursing units at the time of the patient's discharge and taken to the Health Information Management (HIM) Department for processing, including coding. Even in the largest facilities there are less than an average of 250 inpatient discharges a day and most facilities have less than 100 records a day to process. Outpatient records are generated in many locations - emergency rooms, ambulatory surgery centers, diagnostic areas, specialty treatment areas, clinics, off-campus facilities, and others. There are often hundreds or tens of hundreds of records generated daily. In the past, these records or documents may not routinely have been sent to the HIM Department, but kept in satellite record repositories or computer systems. Since each encounter must be coded with a diagnosis and CPT procedure code and since coding should not be done without the medical record, the resulting paper flow issues have been problematic for most hospitals.
There are two alternatives: bring the records to the coders or send the coders to the records.
Transporting coders to the location of the records reduces coder productivity and is costly. There may not be space at remote locations for coders to work or there may not be access to coding software and/or references. Similarly, hiring and/or training coders to work in specific ancillary locations or satellite facilities is costly. These choices will increase the need for coders, who are already in short supply.
A more efficient choice is to utilize technology to bring the medical records to the coders. There are two methods for doing this, currently: an electronic medical record (EMR) or web-based coding.
Electronic Medical Record
Electronic medical record systems can ease this dilemma by storing electronic data and record images in a central repository. With a totally functioning EMR, patient information is available at the source of care and also at the location where coders work. However, most facilities do not yet have electronic records and must deal with paper documents.
With this recently introduced technology, medical record documents can be scanned and transmitted to a secure web site. Coders with access to the web site can view the record images and determine code assignments. The original records can remain at the point of care for on-going treatments and follow-up visits while coding can be completed on a timely basis for bill submission. This technology allows the records to be brought to the coder, wherever the coder is located - even at home.
The same record image can be accessed for both facility and physician coding, eliminating a duplication of effort and a struggle for control of the record that occurs in many facilities. Records can be assigned to coders based on a variety of data elements, including type of record (emergency room, radiology, lab) or payor type.
Web-based coding is safe if the transmissions are encrypted and the images are stored in an encrypted format. Other safeguards that should be present include disabling the print function so that coders (especially those working from home) cannot print
copies of the medical record documents. The record images should not be allowed to reside on the coder's computer. Finally, the software must include an audit trail so that all transactions are recorded and can be monitored.
Not only does web-based coding improve documentation flow to the coder, it also has the potential to improve coding quality. Notes and/or hypertext links allow a coding quality reviewer or supervisor to follow coder's logic in reviewing record because there is an audit trail of the pages that the coder reviewed during the coding process. The coder can add a question to an electronic note "pasted" on a page and send the chart to a supervisor for further review and interpretation. For a coding quality audit, charts can be pre-selected for a review queue and the results of the review are stored.
The web-based technology also allows a facility to utilize outside vendors to handle coding backlogs or to outsource some or all of the coding function. The facility no longer has to pay for travel expenses for coders brought into their HIM department. They don't have to make copies of records and mail or fax them to off-site coders. They simply scan the records to the secure web site and the vendor's coders access the images and code the records. The codes are returned to the facility in less than one day.
Other technology may automate coding at least for straightforward diagnostic procedures (e.g. radiology, cardiology) or minor procedures (e.g. GI lab). A combination of the technologies involved with speech recognition and natural language processing will soon allow dictation to be translated into codes. First, the dictation will be translated into text through a speech recognition process. Next, a natural language processor will process the text to extract the terms describing the procedures and diagnoses. Finally, those terms will be sent through coding and grouping software to produce the ICD-9-CM and CPT-4 codes needed for the bill.
Speech Recognition Technology has had varying degrees of success in the past. However, accuracy is improving due to larger, faster CPUs and smart software that learns more vocabulary and becomes context-sensitive. In other words, it can distinguish between the word "pneumonia" and the phrase "history of pneumonia". The software builds a network of rules that allow it to make reasonable guesses for a word based on the probability of its likelihood to be used with the words around it. However, despite these advancements, speech recognition is not 100 percent accurate. To a large extent, the problem is the English language. It is difficult to distinguish many words, for example, corpse vs. corps, horse vs. hoarse, dilate vs. die late, or even what vs. that.
When using speech recognition, transcriptionists become text editors or physicians may do their own editing. Currently, speech recognition works well for "limited" vocabulary specialties such as radiology, pathology, or the emergency department. Several studies have been done with some remarkable results. One study by Intermountain Health Care showed transcription turn-around time to be 20 min vs. 20 hrs. A Duke University study revealed a 90 percent turn-around time reduction and transcription costs dropped 87 percent.
Natural Language Processing (NLP) is the ability of computers to read and understand words or phrases contained in an ASCII free-text document. The software will extract the words/phrases in transcribed text and process through to encoder/grouper software. Advanced versions of NLP do more than merely search for words. NLP now looks at sentence structure, analyzes words for meaning within a phrase, and considers the meanings of words in conjunction with other words around it.
Natural Language Processing continues to have a problems in determining when to capture a diagnosis or procedure term for coding and when not to. Different rules apply for inpatients and outpatients that complicate this determination. For example, a diagnosis of "probable pneumonia" would be coded as pneumonia for an inpatient, but for an outpatient, the diagnosis would not be coded at all. Instead, outpatient coding must rely on what is known, not on what is suspected. NLP systems also may have problems distinguishing active conditions currently being treated from those that existed in the past and are now only referenced as part of the patient's history.
Here is an example of a short note and the resulting codes produced by a natural language processor. 65yo female; history of right breast cancer seen in the SurgiCenter for biopsy of a breast lesion. Frozen section report shows a benign tumor. Bleeding followed the biopsy. The wound was reopened and bleeders coagulated. The wound was then closed and the patient admitted for observation of post-op bleeding. There was no bleeding recurrence after admission.
Final Dx: Benign neoplasm, left breastThe codes resulting from the Natural Language Processor are
- Diagnosis Codes
- V10.3 History of Breast Cancer
- 998.11 Post-operative Bleeding
- 611.9 Breast Disorder
- 217 Benign Neoplasm, Breast
- V71.9 Observation
- Procedure Codes
- 39.98 Control of hemorrhage
- 85.11 Biopsy of Breast
- 85.0 Mastotomy
- 85.81 Suture of laceration of breast
With the information given, it is not possible to correctly code the note. The code selection would change, depending on whether the note was a summary of the procedure episode or a note related to the reason for observation.
Studies have been done to determine if coding done using NLP us accurate. A 3M study of 996 transcribed dictations from an emergency department (JAHIMA 9/2000) showed the following results:
- 13 percent of the charts were not formatted correctly and rejected by NLP
- 54 percent needed additional manual review because the NLP was unsure of the final code selection
- 33 percent charts were final coded using the NLP
When experienced coders reviewed the automated codes, there was agreement with ICD-9 -CM codes 86 - 90 percent of the time. The differences due to:
- Symptoms coded with related diagnosis
- Differences in 4 th /5 th digits assignments
- Coding of probable diagnoses as actual diagnoses
The coders agreed with Evaluation and Management (E/M) code assignment 80 percent of the time.
Coder productivity increased with the help of NLP and computer generated codes Without NLP, coders averaged 6.32 min/ emergency department chart. With NLP, the average dropped to 3.29 min/chart - a 48 percent improvement. Another recent study showed improvement in coder productivity from 30 - 50 percent (JAHIMA 10/01)
Codes generated through Natural Language Processing are more consistent than codes generated by coders. The same rules and logic are applied each time. Since the codes are based purely on the text in the medical record, NLP-based coding is more compliant than human interpretation, which may read diagnoses or procedures not actually documented, but only inferred.
The demands for coding increase daily with such payment mechanisms as Ambulatory Payment Classifications. Completing the coding on a timely basis is a challenge to most institutions. However, by using technology, these challenges can be overcome.
Speech Recognition Technology: An Outlook for Human-to-Machine Interaction; Erdel, Tim and Crooks, Steve; Journal for Healthcare Information Management ; Summer 2000, 13ff
English is Tough Stuff; Haymont, George; For the Record ; April 2, 2001
The Advances in Automated Coding through Natural Language Processing (NLP); Holbrook, John, MD and Helnze, Daniel T, PhD; 2000 AHIMA
Update on Speech Recognition; Holbrook, John, MD; Advance for Health Information Executives ; March 2000, p. 109ff
Natural Language Processing: A Coding Professional's Perspective; Schnitzer, Gregory L; Journal of AHIMA ; Sep 2000, p 78ff
Can Natural Language Processing Aid Outpatient Coders?; Warner, Homer R; Journal of AHIMA ; Sep 2000, p. 78ff
Speech Recognition Software Can Save Time and Money; Medical Record Briefing ; Mar 2000, p1ff
Source: 2004 IFHRO Congress & AHIMA Convention Proceedings, October 2004Source: library.ahima.org