Relying on Rx Data for Predictive Modeling By Richard Lieberman

Relying on Rx Data for
Predictive Modeling
By Richard Lieberman
Health Data Services, Inc.
Why ACG-Rx?
• What if your health plan doesn’t have
diagnosis codes, but has NDC codes from Rx
transactions?
• Are Rx transactions more reliable than
claims, especially when providers are paid by
capitation?
• Will Rx-based risk adjustment address the
“lag problem” that can hamper predictive
modeling?
Some of the Challenges
Faced
• Drug use is NOT synonymous with
morbidity
• Multiple indications for same drug
– Approved uses
– Off-label uses
• Patterns of practice can directly influence
risk scores
• Complexities of working with roughly
90,000 NDC codes that regularly change.
Examples of the Rx-Morbidity Groups
• Allergy/Immunology / Asthma
• Gastrointestinal/Hepatic / Peptic Disease
• Allergy/Immunology / Chronic Inflammatory
• Genito-Urinary / Acute Minor
• Cardiovascular / Vascular Disorders
•
• Ears, Nose, Throat / Acute Minor
• Endocrine / Bone Disorders
• Endocrine / Diabetes With Insulin
• Endocrine / Diabetes Without Insulin
• Endocrine / Thyroid Disorders
• Female Reproductive / Contraception
• Female Reproductive / Infertility
• Female Reproductive / Pregnancy and
Delivery
• Gastrointestinal/Hepatic / Acute Minor
• Gastrointestinal/Hepatic / Chronic Liver
Disease
• Gastrointestinal/Hepatic / Inflammatory
Bowel Disease
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Infections / Acute MajorGenito-Urinary /
Chronic Renal Failure
Infections / Acute Minor
Infections / HIV/AIDS
Infections / TB
Neurologic / Migraine Headache
Neurologic / Seizure Disorder
Psychosocial / ADHD
Psychosocial / Addiction
Psychosocial / Anxiety
Psychosocial / Depression
Psychosocial / Acute Minor
Psychosocial / Unstable
Skin / Acne
Skin / Acute and Recurrent
Skin / Chronic Medical
www.health-data-services.com
13
Implementing Diagnosis-Based and Rx-Based
Risk Adjustment
• Preparing the data for diagnosis-based risk adjustment is
more complicated
–
Need to eliminate diagnoses from invalid places of service.
•
•
•
•
–
–
–
Diagnostic radiology
Clinical laboratory
Medical supplies/DME
Any service where a clinician did not interact with the patient
Carve-out services
Paid vs. denied services
Temporal issues: measurement period and lag
• Rx-based risk adjustment is easier to implement
–
–
–
All NDCs are used
There are no denied claims (except for reversals)
Most health plans use a single PBM
• Nonetheless, better predictions results when the systems are
used in tandem.
www.health-data-services.com
17
Data Timeframes for Dx-based vs. Rxbased
• Typically, assignment of ACGs, ADGs and EDCs
requires 12 months of medical claims data.
• Allowing for a lag of 3-6 months, this means that
virtually all of the data used for risk adjustment or
predictive modeling is 6-18 months old.
• Pharmacy data does not suffer from the typical 3-6
month time lag.
–
–
Pharmacy claims are adjudicated and paid at the point of sale.
Most PBMs dump data to their health plan customers on a weekly or biweekly
basis.
www.health-data-services.com
18
Number of Months to Maximize Risk
Score
Months of Rx
Data
5
6
7
8
9
10
11
12
Percent of Members
with Maximum Risk
Score
41.9
7.2
7.7
7.4
8.6
9.1
8.8
9.4
www.health-data-services.com
Cumulative
Percent
41.9
49.0
56.8
64.1
72.8
81.9
90.6
100.0
19
Changes in Prevalence of Conditions from Pre- to Post-Disaster:
Conditions Used by Medicare for Capitation Payment
Computation
Johns Hopkins Rx-Morbidity Groups
Prevalence/1,000 members
Pre-Disaster
Post-Disaster
Percent Change
48.5
96.0
+98%
Cardiovascular / Vascular Disorders
Endocrine / Diabetes With Insulin
154.7
59.8
268.1
132.2
+73%
+121%
Endocrine / Diabetes Without Insulin
Gastrointestinal/Hepatic / Chronic Liver
Disease
Gastrointestinal/Hepatic / Pancreatic
Disorder
177.3
296.8
+67%
1.7
2.2
+31%
2.0
4.5
+127%
Genito-Urinary / Chronic Renal Failure
Hematologic / Coagulation Disorders
Infections / HIV/AIDS
Malignancies
1.9
0.3
0.8
20.9
7.7
1.0
3.4
39.6
+310%
+200%
+305%
+90%
Musculoskeletal / Rheumatic Diseases
14.3
31.3
+119%
Neurologic / Parkinson's Disease
15.7
36.4
+132%
Neurologic / Seizure Disorder
Psychosocial / Addiction
75.9
4.1
121.3
9.1
+60%
+122%
Psychosocial / Chronic Unstable
26.4
62.4
+136%
Cardiovascular / Congestive Heart Failure
www.health-data-services.com
20
Changes in Prevalence1 of Mental Illness
Based on Prescription Drugs Coded to RxMGs
Johns Hopkins Rx-Morbidity Groups
Prevalence/1,000 members
Pre-Disaster
Post-Disaster
Percent Change
Psychosocial / ADHD
2.4
7.5
+219%
Psychosocial / Addiction
4.1
9.1
+122%
159.8
181.4
96.7
26.4
252.4
270.6
150.6
62.4
+58%
+49%
+56%
+136%
Psychosocial / Anxiety
Psychosocial / Depression
Psychosocial / Acute Minor
Psychosocial / Unstable
www.health-data-services.com
Benefits of Rx-MGs and RxPM
• More timely “signal”
• Useful building blocks (the Rx-Morbidity
Groups)
• Easier to implement
www.health-data-services.com
22
For More Information
Richard N. Lieberman
Health Data Services, Inc.
www.health-data-services.com
1-866-377-4929
`