User:Little pob/Clinical coding

Medical coding, also clinical coding, diagnostic coding, or coding, is the process of assigning statistical codes from a medical classification to an episode of health care.

Coding requires knowledge of medical terminology and disease processes, attention to detail, and analytical skills. It is most often performed by specialized health informatics staff referred to as medical coders (sometimes clinical coders). However, in some countries and settings coding is performed out by the responsible clinician. With the advent of electronic health records and systems such as SNOMED CT, it is possible for the coding process to be automated.

The diagnosis codes used within coding allow the tracking of diseases and other health conditions, including chronic conditions such as diabetes mellitus and heart disease, and contagious diseases such as norovirus, and the flu. Coded data are used for a variety of applications in medicine, public health and medical informatics, including: statistical analysis of diseases and therapeutic actions, reimbursement, direct surveillance of epidemic or pandemic outbreaks, knowledge-based and decision support systems. Coding for reimbursement is utilised in both the private and social healthcare sectors.

Coding in practice
The basic task of the person coding (the coder) is to translate medical and health care concepts into statistical codes, using a standardized classification such as ICD-10 or DSM-5. Most aspects of health care and outcomes can be coded; inpatient, outpatient episodes, general practitioner visits, population health studies and mortality events can all be coded.

The process of clinical coding can be simplified to three basic phases: abstraction, assignment, and review.

Abstraction
The abstraction phase involves reading the entire record of the health encounter and analysing the information to determine the patient's condition(s), what caused that condition, and any medical or surgical treatment the patient may have had. The information comes from a variety of sources within the medical record, such as clinical notes, laboratory and radiology results, and operation notes.

Assignment
The assignment phase has two parts: finding the appropriate code(s) for the abstraction within the classification; then entering the code into the system being used to collect the coded data. In fact, some countries train their coders that these are two separate steps.

Volume 2 of ICD-10 provides an 8-part break down of the assignment process:
 * 1) Identify the type of clinical statement to be coded; disease, injury or external cause. Then refer to the relevant section of the Alphabetical Index.
 * 2) Locate the lead term in that section of the Index.
 * 3) Read any note that appears under the lead term, and follow its guidance.
 * 4) Read any modifiers after the lead term. These may be enclosed in parentheses (in which case the code number is unaffected), or indented under the lead term (these terms may affect the code number of the parent term). This is done until all the words in the clinical statement have been covered.
 * 5) Carefully follow any “see” and “see also” cross-references within the Index.
 * 6) Verify the suitability of the code number selected by referring to the Tabular List. Any forth character omitted from the index, indicated by a dash in the forth position, would be found at this step. (The further subdivision for use in the supplementary character position, such as those used in chapter XIII, can only be located in Volume 1.)
 * 7) Pay attention to any inclusion or exclusion terms; which may be under the selected code, or at the chapter, block or category heading level.
 * 8) Assign the code.

Review
Reviewing the code set produced from the assignment phase is very important. The coder must ask themselves, "does this code set fairly represent what happened to this patient in this health encounter at this facility." By doing this, clinical coders are checking that they have covered everything that they must, but not used extraneous codes. For health encounters that are funded through a case mix mechanism, the coder might also review the diagnosis-related group to ensure that it does fairly represent the health encounter.

Reimbursement models
Assigned codes and other patient data are processed by grouper software to determine a diagnosis-related group (DRG) for the episode of care, which is used for funding and reimbursement. This process allows hospital episodes to be grouped into meaningful categories, helping us to better match patient needs to health care resources.

Medical billing
Using a common set of published diagnostic and procedures codes, the person coding will translate a patient's episode of care to a list of codes, which will be reported to the health insurance provider. The use of standardized codes, such as those of the International Classification of Diseases or the Common Coding System for Healthcare Procedures, allows insurance providers to map equivalencies across different service providers who may use different terminologies or abbreviations in their written claims forms, and be used to justify reimbursement of fees and expenses. The codes may cover topics related to diagnoses, procedures, pharmaceuticals or topography. The medical notes may also be divided into specialities, for example cardiology, gastroenterology, nephrology, neurology or orthopedics.

Payment by Results
Payment by Results (PbR) is the main reimbursement method used in secondary care within the NHS.

Capitation
With the capitation model a health care provider, such as a hospital, has a contract with a health care commissioner, for example a general practitioner, to provide a certain number of treatment functions over a given period. Within this model the coded data can be used to adjust the payments during any contract renegotiation.

Comparing and analysing data
Generally coding is a concept of modeling reality with reduced effort but with physical copying.


 * Hence the result of coding is a reduction to the scope of representation as far as possible to be depicted with the chosen modeling technology. There will be never an escape, but choosing more than one model to serve more than one purpose. That led to various code derivatives, all of them using one basic reference code for ordering as e.g. with ICD-10 coding. However, concurrent depiction of several models in one image remains principally impossible.


 * Focusing a code on one purpose lets other purposes unsatisfied. This has to be taken into account when advertising for any coding concept. The operability of coding is generally bound to purpose. Inter-referring must be subject of evolutionary development, as code structures are subject of frequent change.


 * Unambiguous coding requires strict restriction to hierarchical tree structures possibly enhanced with multiple links, but no parallel branching for contemporary coding whilst maintaining bijectivity.


 * Spatial depictions of n-dimensional code spaces as coding scheme trees on flat screens may enhance imagination, but still leave the dimensionality of image limited to intelligibility of sketching, mostly as a 3D object on a 2D screen. Pivoting such image does not solve the intelligibility problem.


 * Projections of code spaces as flattened graphs may ease the depiction of a code, but generally reduce the contained information with the flattening. There is no explanation given with many of the codes for transforming from one code system to another. That leads to specialized usage and to limitations in communication between codes. The escape is with code reference structures (as e.g. not existing with SNOMED3).


 * Hierarchical ordering of more than one code system may be seen as appropriate, as the human body is principally invariant to coding. But the dependency implied with such hierarchies decrease the cross referencing between the code levels down to unintelligibility. The escape is with hyper maps that exceed planar views (as e.g. with SNOMED3) and their referring to other codes (as e.g. yet not existing with SNOMED3).


 * Purpose of documenting will be seen as essential just for the validation of a code system in aspects of correctness. However this purpose is timely subordinate to the generating of the respective information. Hence some code system shall support the process of medical diagnosis and of medical treatment of any kind. Escape is with a specialised coding for the processes of working on diagnosis as on working with treatment (as e.g. not intended with SNOMED3).


 * Intelligibility of results of coding is achieved by semantic design principles and with ontologies to support navigating in the codes. One major aspect despite the fuzziness of language is the bijectivity of coding. Escape is with explaining the code structure to avoid misinterpreting and various codes for the very same condition (as e.g. yet not served at all with SNOMED3).

Australia
Australia has a modified version of the ICD called the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM), is applied in all Australian acute health facilities. It is based on the ICD-10 system, published by the World Health Organization, and updated with the Australian Coding Standards (ACS), produced by Australian Classification of Health Interventions (ACHI).

The assigned codes and other patient data, such as age and length of stay, are processed by grouper software to determine a diagnosis-related group (DRG) for the episode of care, which is used for funding and reimbursement. This process allows hospital episodes to be grouped into meaningful categories, helping to create more accurate case mixes to better allocate health care resources.

Germany
In Germany, the coding is performed by clinicians in real time, using SNOMED CT. A background process then cross-maps the SNOMED-CT codes to the German classifications; ICD-10-GM (German Modification) and OPS-301. A data verification process is then undertaken by trained coders.

United Kingdom
For diagnosis coding, the UK's National Health Service uses ICD-10 5th edition. The mandated procedural classification is OPCS Classification of Interventions and Procedures (OPCS-4). Currently primary care activity and findings are captured using clinical terms that are mapped to Read codes. Some of which have a cross-map to the ICD-10 or OPCS-4 classifications.

In the UK's private sector, ICD-9 and a modified version of OPCS-4 are used.

United States
The United States currently use ICD-9-CM. The deadline for switching to ICD-10-CM is October 1, 2015.