MARKET RESPONSE TO FOOD AND DRUG ADMINISTRATION’S SAFETY WARNINGS: A CASE STUDY
USING AN INTERRUPTED TIME SERIES ANALYSIS OF THE MEDICARE DATABASE FOR 2006-2008
Hafiz Oko-osi MS1, Benjamin F. Banahan III PhD1,2, Donna West-Strum PhD1,2, John P. Bentley PhD1,2
1Department
of Pharmacy Administration, School of Pharmacy, University of Mississippi; 2Center for Pharmaceutical Marketing and Management, University of Mississippi
BACKGROUND & OBJECTIVE
Monthly Percentage Utilization of Thiazolidinediones by Patient Group
Poor adherence to Food and Drug Administration (FDA) safety alerts and warnings has the
potential to expose patients to harmful effects of drugs. There is conflicting evidence on the
adequacy of safety warnings to limit drug utilization in patients contraindicated to using
thiazolidinediones (TZDs) (Wilkinson et al. 2004; Shah et al. 2010). Appropriate risk management
should result in decreased use of a product in contraindicated patients while retaining use in
appropriate patients. The objective of this study was to evaluate the impact of FDA safety warning
about the cardiovascular safety of rosiglitazone on the utilization rates of thiazolidinedione oral
anti-diabetes medications among indicated and non-indicated patients.
35%
Percentage of patients
30%
METHODOLOGY
A retrospective analysis of the five percent national sample of Medicare Part D beneficiaries for
years 2006-2008 was conducted using an interrupted time series consisting of a 13-month preintervention period and a 19-month post-intervention period. Beneficiaries with at least one
prescription claim for a TZD anti-diabetic during the study period were included in the sample.
Beneficiaries were classified each month as non-user or rosiglitazone/pioglitazone user based on
TZD possession. Beneficiaries were further classified each month during the study period into
appropriate-use, at risk, and contraindicated groups based on the presence of certain comorbid
conditions using ICD-9-CM diagnoses codes. Rosiglitazone and pioglitazone utilization rates were
calculated each month for the beneficiaries in each appropriateness of use patient group.
Segmented regression analysis was used to examine the effects of the May 2007 FDA safety
warnings about the ongoing review of rosiglitazone’s potential to increase cardiovascular risks on
TZD utilization rates for the different appropriateness of use patient groups.
25%
20%
15%
10%
May 2007 Safety Warning
5%
0%
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Variable
a. Full segmented regression model
Intercept
Time
Intervention
Time after intervention
b. Most parsimonious segmented regression
model
Intercept
Time
Intervention
Time after intervention
Estimate Standard
tError
statistic
P-value
Estimate Standard
tError
statistic
Rosiglitazone
P-value
20.562
0.117
-2.445
-0.874
2.540
0.280
1.100
0.408
8.09
0.42
-2.22
-2.14
< 0.0001
0.6792*
0.0356
0.0421
16.455
0.744
1.247
-0.561
1.433
0.161
1.148
0.220
11.48
4.61
1.09
-2.56
< 0.0001
< 0.0001
0.2869*
< 0.0001
21.506
1.173
18.33
< 0.0001
16.286
0.805
1.539
0.163
10.59
4.93
< 0.0001
<0.0001
-2.421
-0.719
1.078
0.144
-2.25
-5.01
0.0333
< 0.0001
-0.583
0.236
-2.47
Variable
a. Full segmented regression model
Intercept
Time
Intervention
Time after intervention
b. Most parsimonious segmented regression
model
Intercept
Time
Time after intervention
Estimate Standard
tError
statistic
23
24
25
26
27
28
29
30
31
32
% of appropriate-use patients taking Pioglitazone
% of at risk patients taking Rosiglitazone
% of at risk patients taking Pioglitazone
% of contraindicated patients taking Rosiglitazone
% of contraindicated patients taking Pioglitazone
CONCLUSIONS
Full and Most parsimonious segmented regression models predicting the monthly
utilization rate of rosiglitazone and pioglitazone in at risk patients.
Pioglitazone
22
% of appropriate-use patients taking Rosiglitazone
A graph showing the trend in TZD utilization rates by patient groups is presented. The full segmented regression model provides estimates of utilization level and trend before and
after the FDA safety warning for each patient group. The full and most parsimonious segmented regression models reported below were used to estimate the relative difference in
utilization rates for each patient group.
There was an increasing trend in the total utilization rates of thiazolidinediones before the safety warning. Significant decline in drug utilization rates were observed at the end of the
study period for all patient groups on rosiglitazone (relative difference -74.78%, -79.93%, and -90.21% in appropriate-use, at risk and contraindicated patient groups, respectively).
The intervention did not have significant immediate effects on the post-intervention utilization rates of pioglitazone. However, after the intervention, a general decline in utilization of
thiazolidinediones, including pioglitazone, was observed.
Rosiglitazone
21
Month
RESULTS
Full and Most parsimonious segmented regression models predicting the monthly
utilization rate of rosiglitazone and pioglitazone in appropriate-use patients.
20
Full and Most parsimonious segmented regression models predicting the monthly
utilization rate of rosiglitazone and pioglitazone in contraindicated patients.
Pioglitazone
P-value
Estimate Standar
d Error
tstatistic
Rosiglitazone
P-value
19.946
0.204
0.398
-1.200
4.120
0.439
0.998
0.653
4.84
0.47
0.40
--1.86
< 0.0001
0.646*
0.693*
0.078
20.850
0.642
1.287
-0.845
1.490
0.165
1.103
0.288
13.99
3.90
1.17
-3.71
<0.001
<0.001
0.2534*
<0.001
21.607
2.053
10.53
< 0.0001
-0.909
0.209
-4.53
< 0.0001
20.707
0.699
-0.853
1.499
0.159
0.230
13.81
4.40
-3.72
<0.001
<0.001
<0.001
0.0198
Variable
a. Full segmented regression model
Intercept
Time
Intervention
Time after intervention
b. Most parsimonious segmented regression
model
Intercept
Time
Intervention
Estimate Standard
tError
statistic
Pioglitazone
P-value
Estimate Standar
d Error
tstatistic
P-value
19.428
-0.485
-4.618
0.073
1.017
0.119
0.953
0.115
19.11
-4.09
-4.84
0.47
< 0.0001
< 0.0001
< 0.0001
0.6419*
24.790
-0.450
-1.222
0.025
0.670
0.828
0.766
0.099
36.99
-5.43
-1.60
0.25
<0.001
<0.001
0.1222*
0.8030*
19.081
-.0437
-4.598
0.744
0.055
0.941
25.64
-7.94
-4.88
< 0.0001
< 0.0001
< 0.0001
24.856
-0.487
0.554
0.029
44.84
16.83
<0.001
<0.001
The initial warning about rosiglitazone’s cardiovascular safety was effective in
decreasing rosiglitazone’s utilization in the targeted population and thus
appeared to achieve the desired safety effects. However, the safety warning
also had spillover effects by reducing utilization in non-targeted patient groups
and of pioglitazone in the targeted population. This analysis demonstrates the
importance of companies managing emerging risk issues in order to create
market differentiation in use among different patient populations. This tactic,
of appropriately differentiating patient populations, should be considered core
to the ultimate strategic goals of marketing and risk management strategies
when safety issues emerge later in product life cycles.
.
ACKNOWLEDGEMENTS
The work reported was conducted as part of thesis project internally supported by the Center
for Pharmaceutical Marketing and Management at The University of Mississippi. Data
access was approved by the Centers for Medicare and Medicaid Services (CMS) under DUA
21594. The views expressed are those of the authors and do not necessarily reflect those of
CMS or the University of Mississippi.
* Insignificant variables dropped by stepwise elimination resulting in absence in the most parsimonious models
International Society for Pharmacoeconomics and Outcomes Research 17th Annual International Meeting, Washington, DC, USA. June 2-6, 2012.
PCV93