Simulation of ICD-9 to ICD-10-CM transition for family medicine: simple or convoluted?

Abstract

Objectives

The objective of this study was to examine the impact of the transition from International Classification of Disease Version Nine Clinical Modification (ICD-9-CM) to Interactional Classification of Disease Version Ten Clinical Modification (ICD-10-CM) on family medicine and identify areas where additional training might be required.

Methods

Family medicine ICD-9-CM codes were obtained from an Illinois Medicaid data set (113,000 patient visits and $5.5 million dollars in claims). Using the science of networks we evaluated each ICD-9-CM code used by family medicine physicians to determine if the transition was simple or convoluted.¹ A simple translation is defined as one ICD-9-CM code mapping to one ICD-10-CM code or one ICD-9-CM code mapping to multiple ICD-10-CM codes. A convoluted transition is where the transitions between coding systems is non-reciprocal and complex with multiple codes where definitions become intertwined. Three family medicine physicians evaluated the most frequently encountered complex mappings for clinical accuracy.

Results

Of the 1635 diagnosis codes used by the family medicine physicians, 70% of the codes were categorized as simple, 27% of the diagnosis codes were convoluted and 3% were found to have no mapping. For the visits, 75%, 24%, and 1% corresponded with simple, convoluted, and no mapping, respectively. Payment for submitted claims were similarly aligned. Of the frequently encountered convoluted codes, 3 diagnosis codes were clinically incorrect, but they represent only < 0.1% of the overall diagnosis codes.

Conclusions

The transition to ICD-10-CM is simple for 70% or more of diagnosis codes, visits, and reimbursement for a family medicine physician. However, some frequently used codes for disease management are convoluted and incorrect, where additional resources need to be invested to ensure a successful transition to ICD-10-CM.

Introduction

The transition to the International Classification of Disease Version Ten Clinical Modification (ICD-10-CM) will have a huge impact on practicing physicians in the United States.² The expected date for the transition is October 1, 2015. The list of potential diagnosis codes in ICD-10-CM is five times larger than its ICD-9-CM counterpart currently used in practice¹.The American Medical Association (AMA) estimates the cost of the transition to ICD-10-CM is between $83,000-$2 million per physician practice.² A more recent Medical Management Group Association (MGMA) report revealed that the average cost to upgrade/replace practice management systems to use ICD-10 diagnosis codes per Full-Time Equivalent practitioner (FTE) is $10,190.00.³ Additional cost to upgrade/replace Electronic Health Records (EHR) to use ICD-10 diagnosis codes will average $9,979.00 per FTE.³ Previous studies have evaluated a number of different medical specialties in regards to the transition to ICD-10-CM.^4–6 The recent ruling by Center for Medicaid and Medicare Services to not deny any claim due to lack of specificity for the first year during the transition to ICD-10-CM highlights the challenge and potential impact of the new coding system.⁷ To our knowledge, no other studies have evaluated the impact ICD-10-CM will have on the practice of family medicine.

The United States is the last country to transition to ICD-10. The difficulties in making this national transition are multifactorial, but all can be overcome.⁸ Some concerns, as evidenced by a Swiss study, showed that it took up to five years before ICD-10-CM became as accurate as International Classification of Disease version 9 (ICD-9).⁹ In a Canadian study the use of ICD-10-CA (Canadian version with less codes) had a variable impact on quality compared to ICD-9-CM.¹⁰ The objective of this study was to examine the impact of ICD-10-CM on family medicine and identify areas where additional training and preparation might be required. The study was approved by our Institutional Review Board, approval number for exempt status: 2012-0773.

Methods

Overview

Family medicine ICD-9-CM codes were obtained from an Illinois Medicaid data set. Using the science of networks (mathematical algorithms to understand complex networks such as the internet, interstate highway system, and social networks) we evaluated each ICD-9-CM code used by family physicians to determine the relative difficulty of the transition on family medicine practices. Then we took the most frequently used codes and those with complex mappings were evaluated by a number of family medicine physicians for clinical accuracy in the mapping forward. (See ‘Categorization of complex mapping’ later in paper) Financial analysis was conducted to examine the impact of the different transitions.

Data set

Data was culled from all Medicaid patients whose primary care provider was affiliated with the University of Illinois on April 2011; a complete data set of all 2010 bills related to those patients created the database.¹ All physician bills were labeled by physician specialty. The data set was filtered for bills submitted by family medicine physicians. A total of 1,635 ICD-9-CM diagnosis terms were submitted by family medicine physicians, for a total of $5.5 million reimbursement and 113,000 patient visits. A subset of frequently encountered codes (codes used for >25 visits) was created. The visits with these 189 ICD-9-CM codes accounted for 106,726 visits (94%) of all visits and 80% ($4,414,005) of all costs.

Mapping from ICD-9-CM to ICD-10-CM codes

The Center for Medicaid and Medicare (CMS) services created the general equivalent mappings (GEM) a directional mapping from ICD-9-CM to ICD-10-CM.¹¹ CMS also created a separate reverse mapping from ICD-10-CM to ICD-9 CM.¹¹ A previous motif analysis tool was used to map all of the ICD-9-CM diagnosis codes.¹

Categorization of complex mapping

The ICD-9-CM diagnosis codes have previously been categorized for the complexity of transition to ICD-10-CM (Figure 1).¹ The categories are simple and convoluted. Within the simple category are three subcategories: identity, where the ICD-9-CM and ICD-10-CM had a one to one mapping (Figure 1); class-to-subclass where additional data are needed to classify the concept in ICD-10-CM – and subclass-to-class where a number of concepts in ICD-9-CM are merged into a single ICD-10-CM concept. An example of the latter subcategory is when depressive disorder and depressive psychosis (unspecified) are mapped to major depressive disorder, single episode, unspecified (see Figure 1).

Figure 1. Mapping Complexity.

Each ICD-9-CM diagnosis code was categorized into one of the five categories. Panel A. The blue circles represent ICD-9-CM codes and the purple circles represent ICD-10-CM diagnosis codes. The smaller circles represent secondary codes related to the category but have separate analysis as a primary code elsewhere. The arrows between the circles represent the mapping by the GEM files provided by the government. Identity is the category where one code replaces another code. Class-to-subclass is where multiple ICD-10-CM codes were represented by a single ICD-9-CM diagnosis code. Additional documentation or detail will be required. Subclass-to-class is where multiple ICD-9-CM codes will be represented by a single ICD-10-CM code. Convoluted are non-reciprocal and have secondary codes confounding the diagnosis concepts. No mapping is where the GEM file does not provide a mapping to the ICD-10-CM codes. Panel B. Categories represent the percentage of diagnosis codes used in the complete data set in each category.

Simple codes are comprised of the above three categories. Convoluted codes represent transitions between coding systems that are non-reciprocal and have secondary codes confounding the diagnosis concepts. An example of a convoluted code is diabetes with or without complications (controlled or uncontrolled) mapping to various diagnostic codes with or without complications, and vice versa (see Figure 1). Lastly, no mapping is an additional category when the CMS methods do not provide any mapping forward to ICD-10-CM, e.g. pregnancy code (see Figure 1).

Data analysis

Using the science of networks, which leverages the relationships between ICD-9-CM and ICD-10-CM as provided by the GEMs files¹², we mapped the family medicine ICD-9-CM diagnosis codes to the corresponding ICD-10-CM codes and labeled them as: simple, convoluted, and no mapping.¹ The benefit of the science of networks is to reveal complex or convoluted transitions to ICD-10-CM mathematically. Initial analysis of each category involved number of codes, patient visits, and payment made for each unique code and for the more frequently (>25) encountered diagnosis codes. (Figure 2)

Figure 2. Analysis of Family Medicine encounters.

Panel A The first graph analyses all 39,251 encounters with family medicine physicians. The graph of the diagnosis codes counts each diagnosis codes as a single value and categories all of the codes as a percentage of the total number of codes. The visits graph analyses all of the visits for a specific diagnosis code and calculates a percentage of total number of visits. The simple diagnosis codes account for more visits than the convoluted codes. The payments analysis took the payments for each diagnosis code and categorized into the simple or convoluted category for the diagnosis codes and divided by the total amount of reimbursement. Panel B. A secondary analysis was performed on all diagnosis codes used greater than 25 times which included a total number of 26,156 visits. The percentages are near equivalent to the complete representation of visits.

For the frequently encountered diagnosis codes data set (189 codes), the diagnosis codes that were labeled as convoluted (52 codes (28%) - see Figure 2) were evaluated by three family physicians and classified as clinically correct or clinically incorrect. With the disagreements, when two of the three clinicians agreed it was incorrect it was listed. Additional financial analysis was performed on the clinically incorrect codes. (Figure 2)

Results

Of the 1635 diagnosis codes used by family medicine physicians, 70% of them are categorized as simple, 27% are categorized as convoluted, and 3% have no mapping (Figure 2). Of all the 113,000 visits, percentages patient visits of each diagnostic code category were 75%, 24%, and 1% for simple, convoluted and no mapping, respectively (Figure 2). The frequently encountered codes (189) had a similar distribution with 28% of the diagnosis codes and visits being convoluted and 33% of payments being convoluted (Figure 2).

Out of the 189 most frequently used codes, the 53 most frequently encountered convoluted diagnosis codes were evaluated for accuracy of clinical mapping. A total of 16 codes were evaluated by one family medicine physician as being incorrect. However, only 3 were deemed clinically incorrect by at least two physicians (Table 1) – this translates into roughly 5% of the frequently encountered convoluted codes as clinically incorrect. However, only .2% of the overall codes are clinically incorrect when including the complete set of 1635 codes.

Table 1.

Convoluted ICD-9-CM codes with clinically incorrect mappings to ICD-10-CM

ICD-9-CM code	ICD-9-CM Name
*250.02	type II diabetes mellitus [non-insulin dependent type] [NIDDM type] [adult- onset type] or unspecified type, uncontrolled, without mention of complication
*625.9	Unspecified symptom associated with female genital organs
*719.44	Joint pain-hand
V06.8	Need for prophylactic vaccination and inoculation against other combinations of diseases
V70.2	General psychiatric examination, other and unspecified
648.83	Abnormal glucose tolerance of mother, antepartum
919.4	Insect bite, nonvenomous, of other, multiple, and unspecified sites, without mention of infection
719.44	Joint pain-hand
780.79	Other malaise and fatigue
787.91	Diarrhea
292.0	Drug withdrawal
682.0	Cellulitis and abscess of face
V07.31	Need for Prophylactic fluoride administration
789.09	Abdominal pain, other specified site

ICD-9-CM: International Classification of Disease Version Nine Clinical Modification

ICD-10-CM: International Classification of Disease Version Ten Clinical Modification

NIDDM: Non-Insulin Dependent Diabetes Mellitus

^*All ICD-9-CM codes had two family medicine physicians agree the mapping was incorrect

Discussion

When coding in ICD-10-CM the Center for Medicaid and Medicare services recommends reviewing the patient documentation and then selecting ICD-10-CM codes. Use of automatic mapping software systems can lead to problems as identified above. As highlighted in this paper, 70% of the diagnosis codes are relatively easily mapped forward. While many family medicine physicians may not know about the GEM, using previously used ICD-9-CM codes as a frame of reference for what the terms are in ICD-10-CM is strongly recommended as a first step. The concept of convolution helps to highlight only the 27% of family medicine codes where the clinical concepts have changed. Documentation or paradigm shifts will likely be necessary in order to fully accommodate the new coding system in ICD-10-CM. One major challenge with convoluted codes will be disease management reports or other medical reports. The nature of these reports typically provide comparison data from month to month, year to year, or reflect seasonal comparison. Changing the clinical concepts between ICD-9-CM and ICD-10-CM could lead to misleading reports. Detailed analysis about the change will be necessary to identify which reports have to be modified when scrutinizing future data.

The example of a convoluted code (Figure 1) Type 2 DM, not stated as uncontrolled, without mention of complication (250.00, ICD-9-CM) is a great example. This code maps forward and reverse to Type 2 DM without complications (E11.9, ICD-10-CM). The reason 250.00 is convoluted is due to the other associated code Other Specified DM without complications (E13.9, ICD-10-CM). E13.9 only maps backwards to 250.00 as well as to Secondary DM, not stated as uncontrolled, or unspecified (249.00, ICD-9-CM). When comparing the results before and after the transition to ICD-10-CM for diabetes registries, the mapping of E13.9 (ICD-10-CM) might inflate the number of Type 2 DM compared to historical data if the analyst or programmer selected 250.00 (ICD-9-CM) instead of 249.00 (ICD-9-CM). The concept of convoluted reveals this complex relationship where additional analysis is required to ensure successful transition. As many providers are now focusing special payments and incentives based on disease registries, convoluted codes that are tied to registries or disease management need even more attention to ensure minimal disruption.

While subclass-to-class is labeled as easy, some of the implications can have a wide effect on family physicians. For example Depression disorder NEC (311) (Figure 1), is mapped together with Depressive Disorder psychosis unspecified (296.20, ICD-9-CM). However, in the data set, Depression disorder NEC (311, ICD-9-CM) is used 239 times where Depressive Disorder psychosis unspecified (296.20, ICD-9-CM) is used 17 times. If the family medicine physicians use educational tools to map the ICD-9-CM codes to ICD-10-CM a question that remains to be seen is will all family medicine physicians label the 256 visits with the ICD-10-CM diagnosis of Major depressive disorder, single episode, unspecified (F32.9, ICD-10-CM). Or if a physician is searching in ICD-10-CM for the diagnosis via a keyword search for depression or mood, will they select F32.9 (ICD-10-CM) or choose a different diagnosis such as “unspecified mood disorder” (F39, ICD-10-CM)? The challenge with the change in codes and selecting an even more generic description of the patient is the ability for the health system to provide comprehensive services to the patient.

Another possible concern with the transition to ICD-10-CM is the potential increase in time required to code. Experience in Australia and Canada with less complex ICD-10 versions, have demonstrated an increased time to code in ICD-10 even after a year of experience.^13,14 A recent study demonstrated that professional coders could need up to 75% more time to code in ICD-10-CM due to the increased size and complexity of coding.¹⁵ In another study, the professional coders still took 50% more time to code in ICD-10-CM compared to ICD-9-CM after a year of experience.¹⁶ Currently professional coders charge approximately $3.25 per encounter (Personal communication for the exact charge rate for an encounter). If the increase in time is directly reflected in the cost for each coder each encounter would increase by $1.62. If a family medicine doctor has 90 encounters a week working for 50 weeks out of a year, the increase in cost totals $7,312 due to the increased time for the professional coders. Alternatively, the onus to code accurately could be placed on the physician which would not directly increase the cost of care but decrease time available to treat patients which is a much more costly consideration.

The clinically incorrect mappings are relatively few at 0.2% of the codes. An example, the diagnosis code type 2 diabetes mellitus of unspecified type uncontrolled without mention of complication (250.02, ICD-9-CM) with an incorrect mapping to ICD-10-CM will likely have a significant impact on diabetes registries for disease management in family medicine, complicating the concept even further compared to the challenges listed in 250.00 (ICD-9-CM) listed above.

Limitations of this study include: Medicaid data set was collected from only a single state. The claims for the state of Illinois may not be representative of the nation. Also the mapping to evaluate the impact of the transition to ICD-9-CM were provided by CMS. A number of commercial providers have also published transition mapping between ICD-9-CM and ICD-10-CM, which are protected by copyright and legal agreements. The evaluation of the transitions are the opinions of the three family medicine physicians and the complexity of medicine is reflected in their disagreements and how no transition to ICD-10-CM was considered incorrect by all three clinicians.

Complex mapping rates in family medicine are similar to those of pediatrics whose convolution rate was one out of four.⁶ In contrast, family medicine mapping is slightly more complex than oncology with a convolution rate of 18%.⁴

Every family medicine physician and practice have different patient disease burdens. The utilization of specific diagnosis codes is affected by local and regional variations. The good news is that CMS will not deny claims for the first year for lack of specificity.⁷ In order to help prepare for both the transition and eventual denial of claims for lack of specificity, the authors recommend evaluating your commonly used ICD-9-CM diagnosis codes and use one of the many free tools to see what the ICD-10-CM codes are affiliated. The Healthcare Financial Management Associations¹⁷ as well as the American Medical Association² have provided a number of tools and educational materials to help family medicine physicians’ transition to ICD-10-CM. Another tool provided by the Center for Medicare and Medicaid services is a clinical concept list for family medicine (https://www.cms.gov/Medicare/Coding/ICD10/Downloads/ICD10ClinicalConceptsFamilyPractice1.pdf) which provides an initial overview of ICD-10 for family medicine physicians.

In family medicine, .2% of the codes are clinically incorrect they are only related to <1% of the overall visits and < 1% of the overall cost of care delivered within family medicine. With ICD-10-CM offering up to 80,000 unique diagnosis codes, the small percentage of error within this subgroup of diagnosis codes seems quite benign.

Table 2.

Convoluted High Frequency Codes

959.01	Head injury NOS
V07.31	Need for Prophylactic fluoride administration
V04.81	Need for prophylactic vaccination and inoculation against influenza
250.00	type II diabetes mellitus [non-insulin dependent type] [NIDDM type] [adult-onset type] or unspecified type, not stated as uncontrolled, without mention of complication
799.9	Other unknown and unspecified cause of morbidity or mortality
250.02	type II diabetes mellitus [non-insulin dependent type] [NIDDM type] [adult-onsettype] or unspecified type, uncontrolled, without mention of complication
V03.2	Need for prophylactic vaccination with tuberculosis [BCG] vaccine
292.0	Drug withdrawal
V06.1	Need for prophylactic vaccination with combined diphtheria-tetanus-pertussis [DTP] [DTaP] vaccine
314.00	Attention deficit disorder of childhood without mention of hyperactivity
V70.5	Health examination of defined subpopulations
314.01	Attention deficit disorder of childhood with hyperactivity
883.0	Open wound of fingers, without mention of complication
388.70	Otalgia NOS
959.4	Other and unspecified injury to hand, except finger
466.0	Acute bronchitis
V03.82	Need for prophylactic vaccination against Streptococcus pneumoniae [pneumococcus]
518.81	Acute respiratory failure
V05.3	Need for prophylactic vaccination and inoculation against viral hepatitis
558.9	Other and unspecified noninfectious gastroenteritis and colitis
V06.8	Need for prophylactic vaccination and inoculation against other combinations ofdiseases
625.9	Unspecified symptom associated with female genital organs
V70.2	General psychiatric examination, other and unspecified
626.4	Irregular menstrual cycle
789.09	Abdominal pain, other specified site; multiple sites
626.9	Unspecified disorders of menstruation and other abnormal bleeding from femalegenital tract
845.00	Sprain of ankle NOS
648.83	Abnormal glucose tolerance of mother, antepartum
919.4	Insect bite, nonvenomous, of other, multiple, and unspecified sites, without mention of infection
649.13	Obesity complicating pregnancy, childbirth, or the puerperium, antepartumcondition or complication
959.3	Other and unspecified injury to elbow, forearm, and wrist
682.0	Cellulitis and abscess of face
959.5	Other and unspecified injury to finger
682.6	Cellulitis and abscess of leg, except foot
V03.81	Need for prophylactic vaccination against Hemophilus influenza, type B [Hib]
682.7	Cellulitis and abscess of foot, except toes
V03.89	Need for other specified vaccination against single bacterial disease
715.90	Osteoarthrosis, unspecified whether generalized or localized, involvingunspecified site
V04.89	Need for prophylactic vaccination and inoculation against other viral diseases
729.5	Pain in limb
V05.4	Need for prophylactic vaccination and inoculation against varicella
780.60	Fever NOS
V06.4	Need for prophylactic vaccination with measles-mumps-rubella [MMR] vaccine
780.79	Other malaise and fatigue
V06.9	Need for prophylactic vaccination with unspecified combined vaccine
781.0	Abnormal involuntary movements
V70.0	Routine general medical examination at a health care facility
786.09	Other dyspnea and respiratory abnormality
V70.3	Other general medical examination for administrative purposes
787.03	Vomiting alone
V70.9	Unspecified general medical examination
787.91	Diarrhea
719.44	Joint pain-hand