This document describes a project to analyze healthcare fraud using ICD-9 and ICD-10 diagnosis codes. It notes that thousands of ICD-9 codes will directly map to ICD-10 codes, but some codes are prone to fraud and could cost billions if not addressed. The project aims to understand fraud dynamics, generate mock patient data for an ENT specialty, and apply fraud analytics methods to identify conditions and monitor provider variations.
American Statistical Association October 23 2009 Presentation Part 1
ICD-9 to ICD-10 Fraud Risk Assessment and Analytics Project
1. Project Description:
•THOUSANDS of ICD-9 map to ICD-10 Directly! They are in
use for patient treatment over 10 years and more.
• Likelihood of these ICD-9 in 2012 as cancelled ICD-10 by
CMS is quite low in October 2014 and beyond.
•The healthcare fraud exposure by these ICD-9 is in Billions
of $ for patient populations by January 2015.
•There is a very urgent need to understand the
fraud dynamics from these diagnoses and
control the $ leakage NOW; not in October
2014 or later when the ICD-9 to ICD-10
conversion is settled down.
2. •
•Some specialties such as ENT do not change much in
the same patient for a given age. The procedures are
not that expensive and hence conducive to fraud such
as billing without services rendered and even
overbilling that a payer notices too late!
• Several stable diagnoses in Cardiology specialty is
easy target for Billions $ upcoding and downcoding.
• Furthermore, Specialty such as Pediatrics and
OBGYN presents special challenge to Fraud
•Detection. Advance analytics IS needed!!
3. •Mock up 300 lines of data from specialty such as ENT.
•Apply Fraud Analytics Concept resulting in Fraud Conditions.
•The Fraud Analytics Methods and Conditions will be used
prospectively to monitor providers practice variations as well
as utilize them as variables in Predictive Analytics.
4. •478.33 paralysis of vocal chords; bilateral partial
•478.5 other diseases of vocal chord (abces,
cellulitis, granuloma, leukoplakia, chronditis,
singer's node)
•478.74 stenosis of larynx
•478.8 Upper respiratory tract hypersensitivity
reaction, site unspecified
•212.1 benign Neoplasm of respiratory and
Intrthoracic Sinuses
7. 350
300 0%
C
h 250
a 50%
Charge_$
r 200 Mean
g +2 StdDev
150
e -2 StdDev
+3 StdDev
100 -3 StdDev
$ 50
0
1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728293031323334353637
Day Number
8. Name: ENT_High_Low_biller.
Metrics (or data inside this variable): -
A. Total charge $ on two FWA Diagnoses/
total charge $ on all five diagnoses.
B. Total charge $ on two FWA Diagnoses/
total charge $ on Three Non FWA diagnoses.
C. Total number of claim lines on two FWA
Diagnoses/ total claim lines on all five
diagnoses.
9. Variable Name= ENT_High_Low_biller_A=0.46.
ENT_High_Low_biller_B=0.85.
ENT_High_Low_biller_C= 0.44
Conclusion: assuming that the suspected provider is in
the network, 44% of the suspected diagnoses lines
yielded 46% of the revenue from those two Dx for this
provider. This provider likes to bill towards maximum
contract charge $ on those two suspected Diagnoses
regardless of patient gender and age (.46/.44=1.05). A
Medicaid payer needs to stop this Now! Do not take a
chance that CMS will delete them in October 2014.
Even if CMS deletes all those thousands of codes (very
unlikely to happen), BILLIONS will be lost by 1st
January, 2015 (remember FWA is 300 Billion $ industry).
10. Six Sigma approach can’t stops bleeding in Millions of $
because one simple method can’t be Silver Bullet!
However, Advance analytics such as Neural Network and
Decision Tree needs to be applied to stop Billions $ Fraud
because billing pattern is getting complicated every day.
Remember that FWA phrase,” Where there are chronic
patients present, Huge fraud is NEVER absent!!”
There is a misconception that FWA and Disease
Management are two different sides of Cost savings
efforts; that it is doubling of efforts. It is NOT!
I’ll show how Six Sigma and Advance Analytics FWA
variables efficiently help reduce cost in Disease
Management also.
11. Navin Kumar Sinha (2012).
Double Check Consulting.
3744 Pennsylvania Ave, Apt 18, Fremont, CA
94536.
Phone: 952-905-6636.
MS: Statistical Genetics: Urbana-Champaign, IL.
MBA: Utah State University (Decision Sciences).
Experience of saving $50 million by conservative
estimates in Healthcare Industry for Branded
Companies all over USA.