budget impact modeling: appropriateness and determining quality input
DESCRIPTION
Budget Impact Modeling: Appropriateness and Determining Quality Input. C. Daniel Mullins, PhD Professor and Chair, PHSR Dept University of Maryland School of Pharmacy. 4 Key Questions. How can we ensure quality of BIA models?. When is it appropriate to do a BIA? - PowerPoint PPT PresentationTRANSCRIPT
Budget Impact Modeling:Budget Impact Modeling:Appropriateness and Determining Quality Appropriateness and Determining Quality
InputInput
C. Daniel Mullins, PhDC. Daniel Mullins, PhD
Professor and Chair, PHSR DeptProfessor and Chair, PHSR Dept
University of Maryland School of University of Maryland School of PharmacyPharmacy
4 Key Questions4 Key Questions
How can we ensure quality of BIA How can we ensure quality of BIA models?models?
What are criteria for a rigorous BIA?What are criteria for a rigorous BIA?
What data elements are input into a What data elements are input into a BIABIA??
When is it appropriate to do a BIA?When is it appropriate to do a BIA?
- and when is it not?- and when is it not?
Key Question #1Key Question #1
When is it appropriate to do a When is it appropriate to do a BIA?BIA?
- and when is it not?- and when is it not?
Appropriate & Appropriate & InappropriateInappropriate
Short term Short term modelsmodels
Lifetime modelsLifetime models
Payer Payer perspectiveperspective
Patient/providerPatient/provider
Cost-Cost-effectivenesseffectiveness
EffectivenessEffectiveness
Key Question #2Key Question #2
What are criteria for a rigorous BIA?What are criteria for a rigorous BIA?
Criteria for a Rigorous BIA Criteria for a Rigorous BIA ModelModel
Academy of Managed Care Pharmacy (AMCP) Academy of Managed Care Pharmacy (AMCP) Format: Key Elements of a Good ModelFormat: Key Elements of a Good Model
~ Structure~ Structure
~ Data~ Data
~ Outputs~ Outputs
AMCP Checklist for Good AMCP Checklist for Good Models: Models: StructureStructure
~ Transparent Transparent
~ Disease progression model Disease progression model
~ Relevant timeframe Relevant timeframe
~ Appropriate treatment pathwaysAppropriate treatment pathways
~ Good mathGood math
AMCP Checklist for Good AMCP Checklist for Good Models: Models: DataData
~ ClinicalClinical~ Epidemiologic Epidemiologic ~ CostCost~ Quality of LifeQuality of Life
Data quality is Data quality is criticalcritical
AMCP Checklist for Good AMCP Checklist for Good Models: Models: OutputsOutputs
Scientific validityScientific validity~ Published in a quality peer-reviewed journal?Published in a quality peer-reviewed journal?
Face validityFace validity~ Do the results make intuitive Do the results make intuitive sense?sense?
Key Question #3Key Question #3
What data elements are input into a BIA?What data elements are input into a BIA?
Learn by doing: A Case Learn by doing: A Case StudyStudy
A hypothetical case study for aA hypothetical case study for a not so hypothetical new drugnot so hypothetical new drug
- Presentation of the model- Presentation of the model
- A walk through the model- A walk through the model
- Model assumptions- Model assumptions
- Model LimitationsModel Limitations
Overview of the presentation of a Overview of the presentation of a modelmodel
- Take home messages- Take home messages
Decision Tree for Selection of Cost-Effective Agent for Hypertension
ACE
ARB
Beta Blockers
CCB
Diuretics
Mortality
SurvivalMyocardial Infarction
Mortality
SurvivalStroke
Congestive Heart Failure
Transplant
No TransplantRenal Failure
No Event
New drug
Cost-Effective Agent
No Intervention
Mortality
Survival
The CE ratio of each drug category is evaluated against No Intervention in addition to active comparators
Diuretics
Mortality
SurvivalMyocardial Infarction
Mortality
SurvivalStroke
Congestive Heart Failure
Transplant
No TransplantRenal Failure
No Event
Cost-Effective Agent
No Intervention
Mortality
Survival
Mortality
SurvivalMyocardial Infarction
Mortality
SurvivalStroke
Congestive Heart Failure
Transplant
No TransplantRenal Failure
No Event
Mortality
Survival
No Intervention
- Presentation of the model- Presentation of the model
- A walk through the model- A walk through the model
- Model assumptions- Model assumptions
- Model LimitationsModel Limitations
Overview of the presentation of a Overview of the presentation of a modelmodel
- Take home messages- Take home messages
Inputs are entered into the model, Inputs are entered into the model, these are processed and out comes the these are processed and out comes the cost-effectiveness resultscost-effectiveness results
Inputs
Results
The model inputsThe model inputs- Initially 100,000 patients enter the model- Initially 100,000 patients enter the model
- Characteristics of population evaluated in the model - Characteristics of population evaluated in the model
- Event probabilities for each of the possible population - Event probabilities for each of the possible population groupsgroups
evaluated in the modelevaluated in the model
- Persistency rate for each of the drug treatment categories - Persistency rate for each of the drug treatment categories
- Anti-hypertensive drug treatment costs and office visit costs- Anti-hypertensive drug treatment costs and office visit costs
- Initial event treatment costs- Initial event treatment costs
- Annual average treatment costs after - Annual average treatment costs after eventevent (the model runs for 5 years)(the model runs for 5 years)
100,000 patients Patient combination (%)Caucasian event probabilities
African American event probabilities Annual persistency proportions
HTN drug treatment costs and office visit costs Initial event treatment costs Annual average event treatment costs
Inputs
Average event probabilities
Calculation 1
Annual persistence adjusted average event probabilities
Calculation 2
Annual event frequency
Calculation 3
Annual total treatment
costs
Calculation 4
Cumulative costs per event avoided
Results
CalculationCalculation 1 1
Average event probabilities
Annual persistence adjusted average event probabilities
Annual event frequency
Annual total treatment costs
Annual costs per event avoided
100,000 patients
Patient combination (%)
Caucasian event probabilities
African American event probabilities
Annual persistency proportions
HTN drug treatment costs and office visit costs
Initial event treatment costs
Annual average event treatment costs
Inputs Calculation 1 Calculation 2 Calculation 3 Calculation 4 Results
Average event probabilities
Input 70% Caucasian (C) and 30%African American (AA):Input 70% Caucasian (C) and 30%African American (AA):
Calculation done for each Calculation done for each event ievent i
Drug Average Event i ProbabilityDrug Average Event i Probability
PPD,A,Event iD,A,Event i = .7 * P = .7 * PD,C,Event iD,C,Event i + .3 * P + .3 * PD,AA,Event iD,AA,Event i
NI Average Event i ProbabilityNI Average Event i Probability
PPNI,A,Event iNI,A,Event i= .7 * P= .7 * PNI,C,Event iNI,C,Event i + .3 * P + .3 * PNI,AA,Event iNI,AA,Event i
Average event probabilities calculation example Average event probabilities calculation example Calculation done for each drug (D) category and the Calculation done for each drug (D) category and the No Intervention (NI) categoryNo Intervention (NI) category
Calculation 2Calculation 2
Average event probabilities
Annual persistence adjusted average event probabilities
Annual event frequency
Annual total treatment costs
Annual costs per event avoided
100,000 patients
Patient combination (%)
Caucasian event probabilities
African American event probabilities
Annual persistency proportions
HTN drug treatment costs and office visit costs
Initial event treatment costs
Annual average event treatment costs
Inputs Calculation 1 Calculation 2 Calculation 3 Calculation 4 Results
Annual persistence adjusted average event probabilities
Persistence adjusted average event Persistence adjusted average event probabilities calculation example probabilities calculation example Calculation done for each year, since persistence can Calculation done for each year, since persistence can change from year to year change from year to year
Persistence adjusted average event probabilities for year 2 (y2):Persistence adjusted average event probabilities for year 2 (y2):
PPP,Event i,y1P,Event i,y1 = .8 * P = .8 * PD,A,Event iD,A,Event i + .2 * P + .2 * PNI,A,Event iNI,A,Event i
Input for year 2: 80% fully persistent, 20% not persistentInput for year 2: 80% fully persistent, 20% not persistent
Calculation 3Calculation 3
Average event probabilities
Annual persistence adjusted average event probabilities
Annual event frequency
Annual total treatment costs
Annual costs per event avoided
100,000 patients
Patient combination (%)
Caucasian event probabilities
African American event probabilities
Annual persistency proportions
HTN drug treatment costs and office visit costs
Initial event treatment costs
Annual average event treatment costs
Inputs Calculation 1 Calculation 2 Calculation 3 Calculation 4 Results
Annual event frequency
Event frequency (EF) Event frequency (EF) Calculation done for each year, since persistence Calculation done for each year, since persistence change and so does the cohort sizechange and so does the cohort size
# Dy1,Event i = EFy1,Event i * Event i Mortality rate
Number of Event i deaths year 1
# Event i deaths in year 1
Number of Event i survivors in year 1
# Event i survivors in year 1 # Sy1,Event i = EFy1,Event i - # Dy1,Event i
Size of year 2 cohort
Year 2 cohort Y2C = 100,000 - EFy1, total events
Event frequency for year 1, Event i EFy1,Event i = 100,000 * PP,Event i,y1
Event frequency for year 1
Calculation 4Calculation 4
Average event probabilities
Annual persistence adjusted average event probabilities
Annual event frequency
Annual total treatment costs
Annual costs per event avoided
100,000 patients
Patient combination (%)
Caucasian event probabilities
African American event probabilities
Annual persistency proportions
HTN drug treatment costs and office visit costs
Initial event treatment costs
Annual average event treatment costs
Inputs Calculation 1 Calculation 2 Calculation 3 Calculation 4 Results
Annual total treatment costs
Annual total treatment costs Annual total treatment costs Calculation done for each year, since event frequency Calculation done for each year, since event frequency change over time due to the decreasing cohort sizechange over time due to the decreasing cohort size
Year 1 total treatment costs
TCy1,event i =[EFy1,event i * Event i initial costs] +
[100,000 * yearly Drug/Office visit costs]
TCy2,event i =[EFy2,event i * Event i initial costs] +
[Y2C * yearly Drug/Office visit costs] +
[# Sy1,Event i * Year 1 Event i average event treatment costs]
Year 2 total treatment costs
Calculation 5Calculation 5
Average event probabilities
Annual persistence adjusted average event probabilities
Annual event frequency
Annual total treatment costs
Annual costs per event avoided
100,000 patients
Patient combination (%)
Caucasian event probabilities
African American event probabilities
Annual persistency proportions
HTN drug treatment costs and office visit costs
Initial event treatment costs
Annual average event treatment costs
Inputs Calculation 1 Calculation 2 Calculation 3 Calculation 4 Results
Cumulative costs per event avoided
Cumulative costs per event avoidedCumulative costs per event avoided Calculation done for each drug treatment category Calculation done for each drug treatment category evaluatedevaluated
CPEA = [ TCy1, all events, NI - TCy1,all events, drug
treatment]
[#EFy1,all events, NI - #EFy1,all events, drug
treatment]
Cumulative costs per event avoided for a drug treatment category
- The lower the “costs per event avoided” the better
- Presentation of the model- Presentation of the model
- A walk through the model- A walk through the model
- Model assumptions- Model assumptions
- Model LimitationsModel Limitations
- Take home messages- Take home messages
Overview of the presentation of a Overview of the presentation of a model model
Model assumptionsModel assumptions - The baseline event probabilities represents an average American hypertensive population (age, gender, co-morbidities)
- Immediate effect of drug treatment persistency status
- Once patients become non persistent with drug treatment, they stay so
- Linear event treatment costs interpolated from missing data
- Same event survival probability applied to each treatment category
- Same annual event probability applied each model year
- Same annual office visit costs across treatment categories
- Presentation of the model- Presentation of the model
- A walk through the model- A walk through the model
- Model assumptions- Model assumptions
- Model LimitationsModel Limitations
- Take home messages- Take home messages
Overview of the presentation of a Overview of the presentation of a modelmodel
LimitationsLimitations
- Future events modeled by down stream event treatment costs
- Patients with multiple factors are not considered in the model (LVH/diab.)
- Average event treatment costs may not be constant in years after the event
- Partial drug treatment persistency is not considered
- Drug treatment switch is not considered
- Presentation of the model- Presentation of the model
- A walk through the model- A walk through the model
- Model assumptions- Model assumptions
- Model LimitationsModel Limitations
- Take home messages- Take home messages
Overview of the presentation of a Overview of the presentation of a modelmodel
Take Home MessagesTake Home Messages
- Drug A reduces DBP by x mm HG and SPB by y mm Hg
- Drug A provides a favorable safety profile
- Drug A improves patient functioning based on physical domain of ABC
- Drug A reduces down stream event treatment costs
Lessons learned and tricks of the Lessons learned and tricks of the tradetrade
# 1 Be transparent# 1 Be transparent
# 2 Describe limitations (see # 2 Describe limitations (see #1)#1)
# 3 Describe the model in a simple form # 3 Describe the model in a simple form (see #1)(see #1)# 4 Get to the # 4 Get to the pointpoint# 5 Stick to the point# 5 Stick to the point
Key Question #4Key Question #4
How can we ensure quality of BIA How can we ensure quality of BIA models?models?
Testing the qualityTesting the quality
Try to “break the model”Try to “break the model”~ Put in “outlier” valuesPut in “outlier” values~ Does the model “explode”?Does the model “explode”?~ Does the model always give the same Does the model always give the same
result?result?
Test for face validity Test for face validity ~ Do the results make intuitive Do the results make intuitive sense?sense? ~ Do the results seem believable?Do the results seem believable?
Ensuring the qualityEnsuring the quality
Allow for Plan-specific valuesAllow for Plan-specific values~ Do the results reflect Plan demographics?Do the results reflect Plan demographics?~ Do the results reflect Plan costs?Do the results reflect Plan costs?
Consider local practice patternsConsider local practice patterns~ Local prevalence Local prevalence ~ Compare to “standard of care”Compare to “standard of care”~ Use inputs that reflect localUse inputs that reflect local
CostsCosts Hospital length of stayHospital length of stay Physician practicesPhysician practices
Provide transparent inputs and Provide transparent inputs and results so that decision-maker results so that decision-maker cancan
Perform their own assessmentPerform their own assessment
Feel comfortable with Feel comfortable with assumptionsassumptions Feel comfortable with inputsFeel comfortable with inputs
Feel comfortable with Feel comfortable with calculationscalculations Feel comfortable with what’s in Feel comfortable with what’s in thethe
“ “black box”black box”
SummarySummary
Present an overview of your modelPresent an overview of your model~ A picture is worth a thousand wordsA picture is worth a thousand words~ Walk the decision-maker through the analysisWalk the decision-maker through the analysis
BIA should be performed over short to BIA should be performed over short to mid-mid- range time periods – not lifetimerange time periods – not lifetime AMCP guidance focuses on:AMCP guidance focuses on:
~ Structure Structure ~ DataData~ OutputsOutputs
ConclusionConclusion
BIA should reflect the appropriate perspective BIA should reflect the appropriate perspective and what they care about and what they care about
BIA calculations should be transparent andBIA calculations should be transparent and provide insight into change in costs:provide insight into change in costs:
~ Drug CostsDrug Costs~ Total Medical CostsTotal Medical Costs
Make the user interface user friendlyMake the user interface user friendly Allow the decision-maker to see or understandAllow the decision-maker to see or understand what’s in the “black box” what’s in the “black box”