Evaluation of Virginias Preferred Drug List

1
Evaluation of Virginias Preferred Drug List

• Policy and Research Division

August 31, 2005
Department of Medical Assistance Services
2
Presentation Outline
?

• Components of Evaluation For This Report

• PDL Process Movement of Prescriptions

• Preliminary Budget Savings

• Study Design For Assessment Of Health Impact

• PDL Health Impacts

• Conclusions

3
PDL Study Requested By DMAS Director And The
General Assembly

• In 2004, the DMAS director requested a
  comprehensive evaluation of the agencys newly
  administered PDL program.
• Focus of the request
• Program implementation -- movement of claims
  through the system and First Healths management
  of the program and its Call Center
• Program savings in overall pharmacy program
• Health impacts of PDL
• One year later the General Assembly placed
  language in the 2005 Appropriation Act formally
  directing the agency to conduct a study of the PDL

4
Three Important Research Questions Shape This
Review

• Within the context of this issue framework, the
  following specific research questions were
  addressed in the agencys full review of the PDL
• Has the PDL program been implemented in a way to
  ensure a high rate of compliance by physicians
  without adversely affecting patient access?
• Is there evidence that total spending in DMAS
  pharmacy program has been reduced since the PDL
  was established as the linchpin initiative in the
  agencys pharmacy program?
• Is there evidence to suggest that the PDL program
  has adversely impacted patient health outcomes
  for those Medicaid recipients who are switched
  from non-preferred to preferred drugs?

5
Presentation Outline

• Components of Evaluation For This Report

?

• PDL Process Movement of Prescriptions

• Preliminary Budget Savings

• Study Design For Assessment Of Health Impact

• PDL Health Impacts

• Conclusions

6
Drug Claims For This Report Were Selected From
Files Containing Over 26 Million Records And Over
8.5 Million PDL-Eligible Claims
Claims Database (Oct 03 to July 05) 26,881,145
PDL Eligible Claims 8,524,437
Pre-PDL Claims 90 Days Prior to Hard Edit Date
(multiple claims per recipient and drug) 829,875
Post-PDL Claims (Jan 04 to July 05) (multiple
claims per recipient and drug) 6,179,588
Pre-PDL By Prescription Single Claim Per
Recipient Per Drug 420,151
Post-PDL By Prescription Single Claim Per
Recipient Per Drug 1,376,196
7
Claims Flow Analysis Had To Account For Changes
Made To The 2004 PDL Hard Edit (Start) Date Or
Preferred Status For Certain NDC Codes (Drugs) In
2005
of Affected National Drug Codes (NDC)
Hard Edit Date
Preferred Status
Analysis
Used 2004 Hard Edit Date (Verified by FH Staff)
107
Changed
Same
Used 2004 Preferred Status and Considered
Administratively Changed in Flowchart
437
Changed
Same
42
Changed
Changed
Used 2004 Hard Edit Date and Preferred Status and
Considered Administratively Changed in Flowchart
N586
8
Nearly 7 of 10 Prescriptions That Were Written
For Non-Preferred Drugs Prior To The Official
Start of PDL Were Switched To Preferred Drugs
Once The Program Was Implemented
Post-PDL
Change to Preferred Rx 86,143
Pre-PDL
Claim Not Yet Submitted
Recent Pre-PDL Rx
Approved as Non-Preferred 11,682
Non-Preferred Rx 124,088
Walk away
Admin Change 1,142
No Claim Found 25,121
Later Refill
No New Claim 25,121
Total Rx 420,151
No Refill
Remained on Preferred Rx 179,600
Recent Pre-PDL Rx
Preferred Rx 296,063
No New Claim 109,467
No Claim Found 109,467
Admin Change 6,996
9
19 Months Into The Program There Are
Substantially More Recipients Receiving Drugs In
The Post-PDL Period Who Did Not Have Paid RX
Claims In The Pre-PDL Period
Claims Database (Oct 03 to July 05) 26,881,145
PDL Eligible Claims 8,524,437
Pre-PDL Claims 90 Days Prior to Hard Edit Date
(multiple claims per recipient and drug) 829,875
Post-PDL Claims 1 to 6 Weeks After Hard Edit
Date (multiple claims per recipient and
drug) 6,179,588
Pre-PDL By Prescription Single Claim Per
Recipient Per Drug 420,151
Post-PDL By Prescription Single Claim Per
Recipient Per Drug 1,376,196
10
Thus All PDL Compliance Rates In This Study Are
Calculated Without Respect To The Pre-PDL Status
Of Any RX Claims.The Compliance Rate Remains
High
PDL Compliance Rate (January 2004 July 2005)
Status of All Claims Paid After PDL Hard Edits
7
Not on PDL
85 Compliance Rate Needed to Achieve Budget
Savings
93
PDL Drug (Compliance)
Total Claims
1,063,636
Includes only paid claims. .
11
Using This Method Only Minimal Variation In
Compliance Rates Across Select Therapeutic
Classes Is Observed
Central Nervous System
Cardiac Medications
Asthma/ Allergy
Diabetes
Anti-Biotics
Gastro
Analgesics
98
96
94
94
93
91
89
85
Compliance Rate Needed to Achieve Budget Savings
Total Claims
135,872
77,981
285,404
55,723
287,149
112,854
64,049
12
Four Out Of Five Requests For Non-Preferred Drugs
Continue To Be Approved by the FH Call Center
Total (January 2004 to June 2005)
January 2004 to September 2004
October 2004 to June 2005
3
Denied
4
3
(Technical Denials)
Prior Authorization Approved
80
82
81
Physician Agreed To Change to a Preferred Drug
17
16
14
34,606
Total Calls
26,887
61,493
Notes There were also 5 reconsiderations (lt 1
percent) in the October 2004 to June 2005 time
period. Call data indicates that FH
staff are answering phone calls within 28 seconds
on average and the call lengths average
two minutes and 27 seconds.
13
Presentation Outline

• Components of Evaluation For This Report

• PDL Process Movement of Prescriptions

?

• Preliminary Budget Savings

• Study Design For Assessment Of Health Impact

• PDL Health Impacts

• Conclusions

14
The Shift In Market Share Created By The PDL
Continues
Pre-PDL Period
Post-PDL Period
PDL Status of RX Claim
17
11
11
12
16
10
10
10
10
9
10
Non-Preferred Drugs
89
83
91
88
90
90
90
90
84
88
89
43
43
43
43
42
41
Preferred Drugs
57
57
57
57
59
58
July
Aug
Sept
Oct
Nov
Dec
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jan -Aug
Sept
2003
2005
2004
15
The Cost Per Script Continues To Be Below The
Projected Amount Since PDL Implementation
Pre-PDL Period
Post-PDL Period
64
62
Forecasted Cost Per Script
60
58
Actual Cost Per Script
56
54
52
50
July
Dec
June
Dec
June
March
March
Sept
Sept
2003
2004
2005
16
Estimated Savings in The Pharmacy Program Since
The Implementation of The PDL (And More Recently
The MAC) Exceed 35 Million
550
Official Medicaid Forecast for Pharmacy
Expenditures
35.2 million
500
450
Actual Medicaid Expenditures
Millions
400
350
300
FY 2001
FY 2002
FY 2003
FY 2004
FY 2005
17
Presentation Outline

• Components of Evaluation For This Report

• PDL Process Movement of Prescriptions

• Preliminary Budget Savings

?

• Study Design For Assessment Of Health Impact

• PDL Health Impacts

• Conclusions

18
Questions Persist Concerning The Impact Of PDLs
On Recipient Health Outcomes

• The principle criticism leveled at PDL programs
  is that the change in drugs destabilizes patients
  causing a number of adverse reactions
• Critics of the program believe that over time,
  effects of the predicted destabilization in the
  Medicaid program for persons on PDLs will be seen
  in
• Greater medical costs
• More frequent use of emergency departments
• Higher rates of hospitalization
• Longer hospital stays

19
Research Questions For This Analysis

• Net of the influence of other factors, are there
  meaningful differences in the total amount of
  Medicaid spending observed for the PDL and
  non-PDL groups during the follow-up period?
• Are Medicaid spending levels for hospital care
  higher for persons on the PDL after accounting
  for the impact of other factors?
• What, if any, differences are observed in the
  utilization of inpatient hospital care?
• Do PDL recipients utilize emergency departments
  for care at a higher rate than their
  counterparts, after controlling for other
  factors?

20
Quasi-Experimental Design Used To Address The
Question Of PDL Health Impacts
21
First Step In Selection Of Study Groups Was To
Identify Claims That Define Each Group
Based On PDL Hard Edit (Start) Dates Established
From 1-04 through 5-04
Change to Preferred Rx 83,197
Claims Used To ID PDL Program Group
Approved as Non-Preferred 19,275
Non-Preferred Rx 102,472
Claims Used To ID Recipient Comparison Group
No New Claim
Total Claims
Remained on Preferred Rx
Recent Pre-PDL Rx
Preferred Rx
No New Claim
22
The Second Step Was To Identify Unique
Recipients, Determine Whether They Were In The
PDL Or Comparison Group, And Examine Their
Characteristics
Switched to Preferred Drugs 83,197
Multiple Claims (Prescriptions) Per Drug Per
Recipient
Total Claims 102,472
Did Not Switch 19,275
59,802
Single Claim Per Recipient
All Drugs Non-Preferred in Post-PDL Period 491
PDL Group
All Drugs Preferred in Post-PDL Period 6,617
Some Drugs Preferred, Some Non-Preferred in
Post-PDL Period 52,694
Control Group
Excluded
23
There Are Key Differences In The Ages, Race, And
Geographic Location Between The PDL And
Comparison Group
GENDER
RACE
AGE
RESIDENCE
PDL
Female
All Other
Under 65
43
Urban
62
70
49
93
33
Suburban
38
30
51
24
Rural
Male
Black
Over 65
Female
All Other
Under 65
Comparison
36
Urban
63
55
74
63
38
7
37
Suburban
37
45
26
N7,108
Over 65
Male
26
Black
Rural
Note Chi-square values are statistically
significant at lt .05.
24
Key Differences Also Exist In The Medications
Used By Recipients In The PDL Study And
Comparison Group
Rate of Use for Certain Medications
Risk Measure
52
KEY
43
PDL Group
Comparison Group
29
26
19
9
7
3
1.54
1.38
Central Nervous System
Cardiac
Gastro-Intestinal
Asthma/ Allergy
CDPS (Case mix Score)

• Chi-square values are statistically significant
  at .0001. Difference is statistically
  significant lt .05.
• The value of the CDPS in this study is
  understated for persons who have Medicare
  coverage

25
The Third Step Was To Establish The Pre- And
Post-Program Periods Based On Each Recipients
1st Service Date Under PDL
Post Program Period Initiated By Date of 1st RX
Service After Hard Edit (Start) Date With
Maximum Possible Claims Run Out to 5-31-05
Pre Program Period Extends Back 12 Months From
The First RX Service Date
1st Hard Edits 1-04 Date of 1st RX Service
After 1-04
5-30-05
1-31-03
2nd Hard Edits 2-04 Date of 1st RX Service
After 2-04
2-28-03
5-30-05
3rd Hard Edits 5-04 Date of 1st RX Service
After 5-04
5-30-05
5-28-03
26
Extending The Follow-up Period Increased
Attrition In Both The PDL And Comparison Group
Recipient Level
6,617
PDL Group
Period of Follow-up After 1st RX Service Date
6,160
6,160
1,194
739
485
491
485
Comparison Group
274
187
1st PDL RX Service Date
Three Months
Six Months
Nine Months
Twelve Months
27
Presentation Outline

• Components of Evaluation For This Report

• PDL Process Movement of Prescriptions

• Preliminary Budget Savings

• Study Design For Assessment Of Health Impact

?

• PDL Health Impacts

• Conclusions

28
Several Measures Used To Evaluate The Health
Effects Of Virginias PDL

• The following outcome measures were tracked for
  persons in the PDL program
• Total Medicaid spending (excluding waiver and
  long-term care maintenance costs)
• Total Medicaid hospital spending
• Recipient hospitalizations
• Emergency room utilization rates

29
Total Medicaid Spending For The Typical Person On
Preferred Drugs Is Actually Less Than The Amount
Observed For Persons On Non-Preferred Drugs
Annualized Medicaid Spending Levels Per Recipient
Average Spending
6,775
Skewness Levels PDL Group
10.19 Comparison Group 5.32
6060
Median Spending Level
2,570
PDL Group
1,765
Comparison Group
N 436 126
N 436 126
Notes Medicaid expenditures are annualized
based on 9 months of data in the post-PDL period
and are adjusted by the number of
days each recipient was eligible during the
period. The post-PDL period is initiated with
the submission of the 1st prescription
after the programs hard edit date for the
relevant drug. Persons with Medicare coverage
were excluded from this analysis
30
The Typical Person On Preferred And Non-Preferred
Drugs Did Not Have Any Hospital Expenditures
During The First Nine Months After Submission Of
A PDL-Eligible Drug Claim
Annualized Hospital Spending Levels Per Recipient
Skewness Levels PDL Group
15.40 Comparison Group 10.53
Average Hospital Spending
1,699
Median Spending Level
PDL Group
67.52
Comparison Group
0
0
N 436 126
N 436 126
Notes Hospital expenditures are annualized based
on 9 months of data in the post-PDL period and
are adjusted by the number of days
each recipient was eligible during the period.
The post-PDL period is initiated with the
submission of the 1st prescription
after the programs hard edit date for the
relevant drug. Persons with Medicare coverage
were excluded from this analysis
31
A Higher Proportion of PDL Recipients Visited The
Emergency Room In The Nine Month Follow-up Period

Recipients With ..........In The Post-PDL Period
PDL (Study) Group (N 434)
Comparison Group (N116)
Emergency Room Visits
26
Mental Health Hospital Days
Mental Health Hospital Visits
In-Patient Hospital Stays
18
3
3
3
3
.04
.02
Notes Follow-up period is nine months from the
date 1st prescription was written after hard
edits were put in place. Persons who
were eligible in the Pre-PDL period but not in
the post period were excluded. Persons with
Medicare coverage were also excluded.

32
However, There Was No Difference In The Number Of
Visits Made To The Emergency Room. The Typical
Recipient In The PDL And Comparison Group Had
None
Annualized Number Of Emergency Room Visits Per
Recipient
Skewness Levels PDL Group
7.63 Comparison Group 4.51
Average Number of ER Visits
Median ER Visits
5.2
PDL Group
Comparison Group
1.2
0
0
N 433 126
N 433 126
Notes Number of visits are annualized based on
9 months of data in the post-PDL period and are
adjusted by the number of days each
recipient was eligible during the period. The
post-PDL period is initiated with the submission
of the 1st prescription after the
programs hard edit date for the relevant drug.
Persons with Medicare coverage were excluded from
this analysis
33
A Similar Finding Is Observed With Respect To
Days Spent In The Hospital
Annualized Number Of Days Spent In The Hospital
Per Recipient
Skewness Levels PDL Group
13.25 Comparison Group 10.88
Average Number of Days in Hospital
Median Days in Hospital
2.4
PDL Group
Comparison Group
.42
0
0
N 436 126
N 436 126
Notes Number of days are annualized based on 9
months of data in the post-PDL period and are
adjusted by the number of days each
recipient was eligible during the period. The
post-PDL period is initiated with the submission
of the 1st prescription after the
programs hard edit date for the relevant drug.
Persons with Medicare coverage were excluded from
this analysis
34
Attempts To Assess Effects Of PDL On Certain
Outcomes Must Also Recognize The Impact Of Other
External Factors
35
When These Factors Are Accounted For The Total
Amount of Medicaid Spending On Persons Switched
To Preferred Drugs Is Actually Less When Compared
To Those On Non-Preferred Drugs
The Net Impact Of The PDL On Total Medicaid
Spending
Factors Consideredin Payment Model
Predicted Payment Level for Persons On
Non-Preferred Drugs

• Pre-PDL Medicaid Spending
• ?Recipients Case Mix Scores
• ? Age of Recipients
• Race of Recipients
• Type of Residence
• Statistically significant effects

Relative Difference For Persons On Preferred
Drugs
-1098
-1742
Differences not statistically significant
Outliers Out
Outliers In
Notes Estimates produced using OLS regression
model. Dependent variable is total Medicaid
spending per days of eligibility over
a nine month post-PDL period. R2 for model with
outliers is.15. For model without outliers the
R2 is .12.
36
The Difference In Medicaid Spending On Hospital
Care For Persons On Preferred Drugs Is Small And
Statistically Insignificant
The Net Impact Of The PDL On Total Hospital
Spending
Factors Consideredin Payment Model

• Pre-PDL Medicaid Spending
• ? Recipients Case Mix Scores
• ? Age of Recipients
• Race of Recipients
• Type of Residence
• Statistically significant effects

Differences not statistically significant
Relative Difference For Persons On Preferred
Drugs
Outliers In
Outliers Out
28
277

Predicted Payment Level for Persons On
Non-Preferred Drugs
Notes Estimates produced using OLS regression
model. Dependent variable is total hospital
spending per days of eligibility over
a nine month post-PDL period. R2 for model with
outliers is .25 No statistically significant
effects in model without outliers.
37
The Impact For Days Spent In Hospital is Small
And Actually Negative
Factors Consideredin Payment Model
The Net Impact Of The PDL On Total Days In
Hospital
Predicted Number of Days In Hospital For Persons
On Non-Preferred Drugs

• Pre-PDL Hospital Days
• ? Recipients Case Mix Scores
• ? Age of Recipients
• Race of Recipients
• Type of Residence
• Statistically significant effects

-.10
-.17
Outliers Out
Outliers In
Relative Difference For Persons On Preferred
Drugs
Differences not statistically significant
Notes Estimates produced using OLS regression
model. Dependent variable is total days in
hospital per days of eligibility over
a nine month post-PDL period. R2 for model with
outliers is .10 and .06 for model without
outliers.
38
Finally, No Discernable Impact Could Be Found On
Emergency Room Use By Recipients Switched To
Preferred Drugs Relative To Their Counterparts On
Non-Preferred Drugs
The Net Impact Of The PDL On Emergency Room Use
Factors Consideredin Payment Model
Relative Difference For Persons On Preferred
Drugs
Differences not statistically significant

• Recipients Case Mix Scores
• ?Age of Recipients
• Race of Recipients
• Type of Residence
• Statistically significant effects

Outliers In
Outliers Out
1.75
.81

Predicted Days Of Emergency Room Use For Persons
On Non-Preferred Drugs
Notes Estimates produced using OLS regression
model. Dependent variable is total emergency
room visits per days of eligibility
over a nine month post-PDL period. R2 for model
with outliers is .03. For model without outliers
the R2 is .03
39
Conclusions

• After nearly 19 months of program implementation,
  Virginias PDL program continues to gets high
  marks
• PDL compliance rate is high and most changes are
  being made voluntarily
• Patients are not being denied drugs
• The program is saving money for the Commonwealth
  
• Though more research is needed, the program does
  not appear to be adversely impacting the health
  status of persons switched to preferred drugs