University Blog Service - University of Texas at...
1
The Impact of a Value - Based Insurance Design for Chronic and Preventive Medications on Adherence in an Integrated Delivery System Esther J. Yi 1,2 I Zolfaghari Kiumars 1 | Linda Chen 1,2 | Paul J. Godley 1,2 I Michel Jeffrey 1 | Karen L. Rascati 2 1 Baylor Scott and White Health, Temple, TX; 2 University of Texas at Austin College of Pharmacy, Austin, TX BACKGROUND METHODS • Traditional cost-sharing models (i.e. copayments, coinsurance, deductibles etc.) reduce the use of high- value services and address the “Moral Hazard”.1 • Value-based insurance design (VBID) is a form of cost- sharing that improves the ease of access for drugs with high value relative to their costs, simultaneously improving health outcomes and healthcare spending.2 • The peer-reviewed literature supports the ability of copay reductions to increase medication adherence rates, but the effects on clinical outcomes, healthcare utilization and costs remain unclear.2-3 • In April 2016, Baylor Scott & White Health implemented a copay reduction program for select chronic and preventive medications. OBJECTIVES • To discuss the use of Interrupted Time Series study designs and Segmented Regression analyses when evaluating policy interventions. • To evaluate the impact of a copay reduction program on adherence rates, clinical outcomes and costs for specific chronic and preventive medications. • BSWH is a non-profit, integrated delivery system (IDS) in Central and North Texas that includes a network of 48 acute care hospital sites, more than 900 patient care sites, more than 6,000 active physicians, and owns and operates 24 retail pharmacies in North and Central Texas. LIMITATIONS CONCLUSIONS REFERENCES 1. Homedes, N. and A. Ugalde, Improving access to pharmaceuticals in Brazil and Argentina. 2006: Oxford University Press in association with The London School of Hygiene and Tropical Medicine. 2. Leyva-Flores R, Brofman M, Erviti-Erice J. 2000. Simulated clients in drugstores: prescriptive behaviour of drugstore attendants. Journal of Social and Administrative Pharmacy 17: 151–8. 3. Cavagnero E, Carrin G, Xu K, Aguilar Rivera AM. Health financing in Argentina: An Empirical Study of Health Care Utilization and Health Care Expenditure. WHO. Geneva. 2006. • Limited generalizability to population outside of selected cohort. • Segmented regression analyses assume linearity in each segment. • A need for further evaluation of the cost and clinical outcomes to understand the full impact and the added value from the policy intervention. Special thanks to: Dr. Paul Godley Dr. Karen L. Rascati Presented at: ISPOR 2018 Conference 19-23 May 2017 – Baltimore, Maryland Esther Jihee Yi University of California at San Francisco School of Pharmacy, Pharm.D. Baylor Scott and White Health, HEOR Post-Doctoral Fellow University of Texas at Austin, M.S. Candidate SETTING Dec 31, 2017 Jan 1, 2014 Figure 1: Study Design Target Population: • Patients > 18 with at least one pharmacy claim for selected medication during study period • Commercially insured patients at BSWH with reduced copayments Select Medications with Reduced Copayments: • 3 disease states: Diabetes, COPD, Anticoagulation • 10 AHFS Drug classes (see results) Adherence: • Monthly proportion of days covered (PDC) Study Period: Jan 1, 2014 to Dec 31, 2017 Intervention of VBID April 1, 2016 Pre-Intervention Jan 1, 2014 – Mar 31, 2016 Post-Intervention May 1, 2016 – Dec 31, 2017 Design • Retrospective cohort study using an interrupted time series design and segmented regression analysis to evaluate the impact of the intervention on adherence, clinical outcomes and cost. • 10 AHFS therapeutic classes with brand-only medications from 3 disease states were selected for the copay reduction program. • Pharmacy claims data (CY 2014 - 2017) from a cohort of commercially insured patients within the integrated delivery system’s virtual data warehouse were obtained to determine medication adherence. • Medication adherence was calculated via monthly proportion of days covered (PDC): the number of days’ supply of a medication within the month divided by the number of calendar days in that month. RESULTS Table 1. Characteristics of Patients in the Study Sample, by Condition Diabetes COPD Anticoagulation Number of patients 202 1317 456 Age, Mean (SD) 53.5 (13.6) 44.9 (14.1) 53.0 (10.2) Female, no. (%) 102 (50.5) 886 (67.3) 260 (57.0) White non-Hispanic (%) 74 (36.6) 428 (32.5) 137 (30.0) Black (%) 5 (2.5) 42 (3.2) 12 (2.6) Asian (%) 3 (1.5) 4 (0.3) 3 (0.66) Hispanic (%) 6 (3.0) 29 (2.2) 24 (5.3) Table 3. Model parameter estimates, and P-values GLP-1 agonists Corticosteroids + B2 agonists Coefficient P-value Coefficient P-value Baseline 0.7859 <0.0001 0.5913 <0.0001 Pre-trend -0.0054 0.0929 -0.0021 0.0038 Level -0.3637 0.0239 -0.1398 0.0039 Post-trend 0.0139 0.0104 0.0062 <0.0001 Chart 1. Monthly medication adherence: anticoagulation Chart 2. Monthly medication adherence: diabetes Chart 3. Monthly medication adherence: COPD Table 5. Model parameter estimates and P-values, COPD Coefficient P-value Baseline mean PDC Anticholinergics 0.5729 <0.0001 Corticosteroids + Beta 2-Agonists 0.5945 <0.0001 Corticosteroids 0.5567 <0.0001 Time 0.0043 0.1698 Corticosteroids + Beta 2-Agonists -0.0065 0.0440 Corticosteroids -0.0047 0.1789 Intervention -0.2481 0.2091 Corticosteroids + Beta 2-Agonists 0.1102 0.5877 Corticosteroids 0.3234 0.1407 Time after intervention 0.0068 0.2819 Corticosteroids + Beta 2-Agonists -0.0006 0.9232 Corticosteroids -0.0077 0.2740 Table 4. Model parameter estimates & P-values, diabetes Coefficient P-value Baseline mean PDC DPP-4 (Reference) 0.8080 <0.0001 DPP-4 + Biguanides 0.7750 <0.0001 GLP-1 0.7706 <0.0001 SGLT2 0.7605 <0.0001 Time -0.0004 0.7614 DPP-4 + Biguanides 0.0033 0.1650 GLP-1 -0.0042 0.2136 SGLT2 0.0003 0.8973 Intervention -0.0131 0.8724 DPP-4 + Biguanides 0.1592 0.3089 GLP-1 -0.3089 0.0759 SGLT2 -0.0661 0.6015 Time after intervention 0.0013 0.6256 DPP-4 + Biguanides -0.0073 0.1451 GLP-1 0.0109 0.0622 SGLT2 0.0019 0.6433 Table 3. Model parameter estimates and P-values, FXa Coefficient P-value Baseline mean PDC 0.6843 <0.0001 Time 0.0027 0.2357 Intervention 0.0789 0.4244 Time after Intervention -0.0023 0.5139 Table 2. Monthly PDC* PDC, Mean (SD) AHFS Drug Class n Pre Post Δ Adherence Anticholinergics 56 0.73 (0.09) 0.72 (0.10) -0.10 Beta 2-Agonists 7 0.46 (0.27) 0.62 (0.14) Corticosteroids + Beta 2-Agonists 818 0.59 (0.02) 0.62 (0.03) 0.03* Corticosteroids 436 0.56 (0.06) 0.60 (0.05) 0.04* Direct FXa Inhibitors 202 0.79 (0.06) 0.81 (0.03) 0.02 DPP-4 Inhibitors 186 0.84 (0.04) 0.86 (0.02) 0.02 DPP-4 + Biguanides 62 0.84 (0.06) 0.84 (0.04) 0.00 GLP-1 Agonists 68 0.72 (0.09) 0.76 (0.06) 0.04 SGLT-2 Inhibitors 132 0.78 (0.05) 0.83 (0.05) 0.05* SGLT2 + Biguanides 8 0.62 (0.27) 0.70 (0.12) • The value-based insurance design produced an intended effect on medication adherence. • Post-period average quarterly PDCs increased by an average of 3.4% from the pre-period. • For both corticosteroids + Beta-2 agonists and GLP-1 inhibitors, we observed a statistically significant increase in the monthly PDC after the intervention of about 1-2%, respectively. • Future analysis on cost and clinical outcomes will provide greater insight on the impact of the value-based insurance design.
Transcript of University Blog Service - University of Texas at...
The Impact of a Value-Based Insurance Design for Chronic and Preventive Medications on Adherence in an Integrated Delivery SystemEsther J. Yi1,2 I Zolfaghari Kiumars1 | Linda Chen1,2 | Paul J. Godley1,2 I Michel Jeffrey1 | Karen L. Rascati2
1Baylor Scott and White Health, Temple, TX; 2University of Texas at Austin College of Pharmacy, Austin, TX
BACKGROUND METHODS
• Traditional cost-sharing models (i.e. copayments,
coinsurance, deductibles etc.) reduce the use of high-
value services and address the “Moral Hazard”.1
• Value-based insurance design (VBID) is a form of cost-
sharing that improves the ease of access for drugs with
high value relative to their costs, simultaneously
improving health outcomes and healthcare spending.2
• The peer-reviewed literature supports the ability of
copay reductions to increase medication adherence
rates, but the effects on clinical outcomes, healthcare
utilization and costs remain unclear.2-3
• In April 2016, Baylor Scott & White Health
implemented a copay reduction program for select
chronic and preventive medications.
OBJECTIVES
• To discuss the use of Interrupted Time Series study
designs and Segmented Regression analyses when
evaluating policy interventions.
• To evaluate the impact of a copay reduction program
on adherence rates, clinical outcomes and costs for
specific chronic and preventive medications.
• BSWH is a non-profit, integrated delivery system (IDS) in Central and North Texas that includes a network of 48 acute care hospital sites, more than 900 patient care sites, more than 6,000 active physicians, and owns and operates 24 retail pharmacies in North and Central Texas.
LIMITATIONS CONCLUSIONS
REFERENCES
1. Homedes, N. and A. Ugalde, Improving access to pharmaceuticals in Brazil and Argentina. 2006: Oxford University Press in association with The London School of Hygiene and Tropical Medicine.
2. Leyva-Flores R, Brofman M, Erviti-Erice J. 2000. Simulated clients in drugstores: prescriptive behaviour of drugstore attendants. Journal of Social and Administrative Pharmacy 17: 151–8.
3. Cavagnero E, Carrin G, Xu K, Aguilar Rivera AM. Health financing in Argentina: An Empirical Study of Health Care Utilization and Health Care Expenditure. WHO. Geneva. 2006.
• Limited generalizability to population outside of selected
cohort.
• Segmented regression analyses assume linearity in each
segment.
• A need for further evaluation of the cost and clinical
outcomes to understand the full impact and the added
value from the policy intervention.
Special thanks to:
Dr. Paul Godley
Dr. Karen L. Rascati
Presented at:
ISPOR 2018 Conference
19-23 May 2017 – Baltimore, Maryland
Esther Jihee Yi
University of California at San Francisco School of Pharmacy, Pharm.D.
Baylor Scott and White Health, HEOR Post-Doctoral Fellow
University of Texas at Austin, M.S. Candidate
SETTING
Dec 31, 2017Jan 1, 2014
Figure 1: Study Design
Target Population:• Patients > 18 with at least one pharmacy claim for selected
medication during study period• Commercially insured patients at BSWH with reduced
copaymentsSelect Medications with Reduced Copayments:• 3 disease states: Diabetes, COPD, Anticoagulation• 10 AHFS Drug classes (see results)Adherence: • Monthly proportion of days covered (PDC)
Study Period: Jan 1, 2014 to Dec 31, 2017
Intervention of VBID April 1, 2016
Pre-InterventionJan 1, 2014 – Mar 31, 2016
Post-InterventionMay 1, 2016 – Dec 31, 2017
Design
• Retrospective cohort study using an interrupted time
series design and segmented regression analysis to
evaluate the impact of the intervention on adherence,
clinical outcomes and cost.
• 10 AHFS therapeutic classes with brand-only
medications from 3 disease states were selected for the
copay reduction program.
• Pharmacy claims data (CY 2014 - 2017) from a cohort of
commercially insured patients within the integrated
delivery system’s virtual data warehouse were obtained
to determine medication adherence.
• Medication adherence was calculated via monthly
proportion of days covered (PDC): the number of days’
supply of a medication within the month divided by the
number of calendar days in that month.
RESULTS
Table 1. Characteristics of Patients in the Study Sample, by Condition
Diabetes COPD Anticoagulation
Number of patients 202 1317 456
Age, Mean (SD) 53.5 (13.6) 44.9 (14.1) 53.0 (10.2)
Female, no. (%) 102 (50.5) 886 (67.3) 260 (57.0)
White non-Hispanic (%) 74 (36.6) 428 (32.5) 137 (30.0)
Black (%) 5 (2.5) 42 (3.2) 12 (2.6)
Asian (%) 3 (1.5) 4 (0.3) 3 (0.66)
Hispanic (%) 6 (3.0) 29 (2.2) 24 (5.3)
Table 3. Model parameter estimates, and P-values
GLP-1 agonists Corticosteroids + B2 agonists
Coefficient P-value Coefficient P-value
Baseline 0.7859 <0.0001 0.5913 <0.0001
Pre-trend -0.0054 0.0929 -0.0021 0.0038
Level -0.3637 0.0239 -0.1398 0.0039
Post-trend 0.0139 0.0104 0.0062 <0.0001
Chart 1. Monthly medication adherence: anticoagulation
Chart 2. Monthly medication adherence: diabetes
Chart 3. Monthly medication adherence: COPD
Table 5. Model parameter estimates and P-values, COPD
Coefficient P-value
Baseline mean PDC
Anticholinergics 0.5729 <0.0001
Corticosteroids + Beta 2-Agonists 0.5945 <0.0001
Corticosteroids 0.5567 <0.0001
Time 0.0043 0.1698
Corticosteroids + Beta 2-Agonists -0.0065 0.0440
Corticosteroids -0.0047 0.1789
Intervention -0.2481 0.2091
Corticosteroids + Beta 2-Agonists 0.1102 0.5877
Corticosteroids 0.3234 0.1407
Time after intervention 0.0068 0.2819
Corticosteroids + Beta 2-Agonists -0.0006 0.9232
Corticosteroids -0.0077 0.2740
Table 4. Model parameter estimates & P-values, diabetes
Coefficient P-value
Baseline mean PDC
DPP-4 (Reference) 0.8080 <0.0001
DPP-4 + Biguanides 0.7750 <0.0001
GLP-1 0.7706 <0.0001
SGLT2 0.7605 <0.0001
Time -0.0004 0.7614
DPP-4 + Biguanides 0.0033 0.1650
GLP-1 -0.0042 0.2136
SGLT2 0.0003 0.8973
Intervention -0.0131 0.8724
DPP-4 + Biguanides 0.1592 0.3089
GLP-1 -0.3089 0.0759
SGLT2 -0.0661 0.6015
Time after intervention 0.0013 0.6256
DPP-4 + Biguanides -0.0073 0.1451
GLP-1 0.0109 0.0622
SGLT2 0.0019 0.6433
Table 3. Model parameter estimates and P-values, FXa
Coefficient P-value
Baseline mean PDC 0.6843 <0.0001
Time 0.0027 0.2357
Intervention 0.0789 0.4244
Time after Intervention -0.0023 0.5139
Table 2. Monthly PDC*
PDC, Mean (SD)
AHFS Drug Class n Pre Post Δ Adherence
Anticholinergics 56 0.73 (0.09) 0.72 (0.10) -0.10
Beta 2-Agonists 7 0.46 (0.27) 0.62 (0.14)
Corticosteroids + Beta 2-Agonists 818 0.59 (0.02) 0.62 (0.03) 0.03*
Corticosteroids 436 0.56 (0.06) 0.60 (0.05) 0.04*
Direct FXa Inhibitors 202 0.79 (0.06) 0.81 (0.03) 0.02
DPP-4 Inhibitors 186 0.84 (0.04) 0.86 (0.02) 0.02
DPP-4 + Biguanides 62 0.84 (0.06) 0.84 (0.04) 0.00
GLP-1 Agonists 68 0.72 (0.09) 0.76 (0.06) 0.04
SGLT-2 Inhibitors 132 0.78 (0.05) 0.83 (0.05) 0.05*
SGLT2 + Biguanides 8 0.62 (0.27) 0.70 (0.12)
• The value-based insurance design produced an intended effect on medication adherence.
• Post-period average quarterly PDCs increased by an average of 3.4% from the pre-period.
• For both corticosteroids + Beta-2 agonists and GLP-1 inhibitors, we observed a statistically
significant increase in the monthly PDC after the intervention of about 1-2%, respectively.
• Future analysis on cost and clinical outcomes will provide greater insight on the impact of the
value-based insurance design.
