Economic and Outcome Studies

Question 0

Examples of patient outcomes:

Answer 0

Economic 
Clinical
 humanistic 
Outcomes 
(ECHO)

Question 1

Definition of outcomes research

Answer 1

Attempt to identify, measure, as evaluate end result of health care services. Compares competing options in routine care settings

Question 2

Example of humanistic outcomes

Answer 2

Health related quality of life

Satisfaction with health care

Question 3

Pharmacoecomonic studies

Answer 3

Includes economic and clinical/humanistic outcome evals and comparisons

Question 4

Efficacy vs effectiveness

Answer 4

Efficacy : randomize controlled trials

Effectiveness: outcomes studies

Efficacy advantages:
Needed for FDA approval, higher internal validity (more sure of cause and effect, less confounding)
Less chance for baseline diff
Higher pt adherence (real or artificial?)

Effectiveness data advantage

• higher external validity (generalize able)
• broader range of pts
• can follow for longer time frame
• larger sample size
• pharmacoeconomic studies included

Question 5

Statistic and value suggesting heterogeneity may be a problem

Answer 5

I-squared value of 50% or more

Question 6

Cost minimization analysis

Answer 6

Assume benefits to be equivalent so lowest cost in $ best

Advantage: simple, no outcomes measures needed
Disadvantage:
- most things have different outcomes so not many uses in real life
- strength depends on strength of evidence that tx equivalent- FDA vs RCT vs “expert opinion”
consequences not measured so some say not true pharmacoeconomic study

Question 7

Cost effectiveness analysis

Answer 7

Measure in Natural units eg life years gained , mm Hg, mmol/L glucose, PANNS score

Advantage: clinician familiar with measuring clinical outcomes
FDA has standards for clinical outcome measurement
Do not have to place dollar value on clinical outcomes

Disadvantage: can only compare options with same type of clinical outcome
Only one clinical outcome at a time can be measured
Metal health outcomes hard to measure

Cost effectiveness grid see p 255

Question 8

Cost benefit analysis

Answer 8

Cost - $
Outcome - Also converted to $ so better outcomes can be associated with higher outcome $$ even if there are higher $$ costs initially
- input and output are $ units - if benefit: cost >1 => cost beneficial

Advantage: can compare tx with different outcomes , and can determine if added benefits outweigh costs

Disadvantage: difficult to put $$ on outcomes, and disagreement on how to value these outcomes

Question 9

Cost utility analysis

Answer 9

QALY
COst utility analysis is cost per outcome of QALY
- some view it as a subset of CEA
Advantage: includes pt preferences
Single measure to incorporate morbidity and mortality
Allows comparisons across different options
Disadvantage: time consuming, results vary and depends on quality measure and instrument

Question 10

Perspective on costs

Answer 10

Payer - insurance co
Provider - hospital or clinic or possibly ACA
Society - includes non medical costs eg homeless shelters, reduced productivity etc

Question 11

Types of costs

Answer 11

Direct medical
Direct non medical
Indirect costs
Tangible costs

Question 12

Direct non-medical costs

Examples and definitions

Answer 12

Law enforcement
Shelters
Jails
Subsidies for housing

Question 13

Indirect costs

Examples and definition

Answer 13

Unemployment
Reduced productivity at work
Premature mortality eg suicide
Caregiver reduced productivity
Level of employment

Question 14

Decision analysis

Answer 14

• systematic quantitative approach for assessing relative value of one or more decision options
• usually depicted by a graph caled a decision tree
  Steps of decision :
  1) identity specific decision - from whose perspective, define options, one what time period
  2) draw decision structure structure over time: boxes are choice nodes (drug a vs drug b), circles represent chance nodes (ADR or no ADR etc)
  Triangles are termination nodes (death, exit plan etc)
  3) assess probabilities
  
  • use past literature, experts judges etc
  • use reasonable ranges for uncertain prob (see sensitivity analysis)
  • 4) determine value of each outcome - must be same for each option eg QALY vs QALY or success rate vs success rate
    5) conduct sensitivity analysis
  • choose values or probabilities most uncertain or where small difference has a big impact
    Use reasonable ranges. Recalculate using the ranges and determine the degree of sensitivity

Question 15

If results of a PE analysis change based on different assumptions - sensitivity analysis - is the result considered sensitive or insensitive?

Answer 15

If results don’t change - insensitive - more robust/ reliable

If results change based on sensitivity analysis - more sensitive - use much more caution in applying and interpreting results

Question 16

One way sensitivity analysis

Answer 16

Only changes assumptions of one variable at a time while holding the others constant

Question 17

Systematic review vs meta analysis

Hint: many systematic reviews use meta analysis

Answer 17

Systematic review: collate empirical evidence with pre-specified eligibility criteria to answer a specific research question

Meta analysis: statistical technique to combine results of previous studies to estimate effect size when smaller studies are pooled. Larger studies weigh more heavily into outcomes than smaller ones

Question 18

Meta analysis transparency tools

Answer 18

QUOROM

PRISMA

Question 19

Steps in meta analysis

Answer 19

1) identify studies
- medline emabse etc
2) define eligibility
- specific and objective
3) extract info
- collect data needed
- use second reviewer for all these steps
4) analyze statistically
- see example p 263-3

Question 20

Test for heterogeneity

Answer 20

Overall result should depend in the intervention and not other variables - use stats to determine if type of funding, type of participants, dose, etc)
- variance higher than that applicable to chance alone = heterogeneity
- two terms:
Cochran Q
I squared - inconsistency measure - measures the variation percent due to heterogeneity rather than chance - >50% implies may be too much heterogeneity

Question 21

Test for publication bias

Answer 21

Funnel plot
Compares the effect (function of risk ratio or risk difference) by precision (sample size or standard error)
- pub bias-> small studies may show larger estimates of effect size
If plot asymmetrical - pub bias may exist - see p 267
- controversy over using this - may be other reasons than pub bias that smaller studies have more effect - different pt Populations, different dose, etc