HCMG EXAM QUESTIONS AND ANSWERS LATEST
EDITION ALREADY GRADED A+
UNCERTAINTY AND THE WELFARE ECONOMICS OF MEDICAL CARE- Kenneth Arrow
Research question: How do the conventional principles of welfare economics apply to the healthcare
sector, given the unique characteristics of medical care as an economic good?
Key findings: Healthcare industry differs from standard economic competitive models
information asymmetry: patients have to trust professionals
Market failure: uncertainty, difficulty in assessing quality, moral hazard, demand unpredictable
Role of health insurance: can mitigate inefficiencies and address uncertainty by spreading risk
Public vs private position: need for government intervention
Problems with insurance:administrative costs, predictability, gaps in coverage, trust, licensing
The Economics of Moral Hazard: Comment - Pauly
Goes against Arrow- governemnt creating insurance can lead to inefficiency
People prefer unfair insurance to self insurance- as long as premiums are not too unfair
Prisoner's dilemma leads to moral hazard
deductibles and coinsurance as solution
Even if all individuals are risk averters, some uncertain medical care expenses will not and should not be
insured in an optimal situation
Insurance should exist where demand is inelastic, random , where individuals have risk aversion
Commercial insurance (mandatory) is wrong; everyone has different tastes → different insuranc
Adjustment for Baseline Characteristics in Randomized Clinical Trials- Holmberg, Andersen
Adjusting for baseline (/ stratification) may improve statistical efficiency if they are prognostic and if
stratified randomization is used
should be prespecified, post hoc selection leads to bias
Adjusting can be too complex if too many variables, Stratification can be biased if too many strata, non
prognostic variables can decrease precision
Authors included stratification variable (enrollment site) and 3 pre specified baseline characteristics
, (age, type of admission, organ score)
testing sodium solution on 90 day survival- no effects
Sequential, Multiple Assignment, Randomized Trial Designs Kidwell, Almirall
A SMART is a type of multistage, factorial randomized trial, in which some or all participants are
randomized at 2 or more decision points (treatment A or B), can answer multiple questions
necessary when the optimal sequence of interventions differs among patients
Limitations: no definitive evidence of the effectiveness of an intervention, a SMART is generally not an
adaptive randomized trial, it is common to conflate the variables
The trial: 2-stage, adaptive telecare intervention to treat psychiatric disorders in underserved, rural
settings. Patients randomized to direct vs indirect telecare in stage 1. At the end of stage 1 (month
6),those not engaged were rerandomized in stage 2 to receive a telephone call vs not.
Found no effect for direct telcare, and no effect for phone calls for those not engaged in telcare.
Missing Data: How to Best Account for What Is Not Known- Newgard, Lewis
many methods for handling missing values can yield biased results & decrease study power
complete case analysis, last observation carried forward, baseline observation carried forward, mean
value imputation, random imputation
"informative" when the absence of a value indicates something about what it should be.
MCAR: prob of being missing is unrelated to all observed and unobserved patient characteristics.
MAR: observed values can "explain" which values are missing predict what it would be
MNAR: missing values are dependent on unobserved or unknown factors (most problematic)
Simple imputation methods are "naive" - can introduce bias, and artificially increase precision
limitation of complete case analysis: bias and reduced sample size, reduced study power.
With more missing data, mean value imputation results in greater bias, artificial precision
Random number imputation fails to use observed values to inform the selected estimate.
LOFC assumes that the final outcome does not change from the last observed value.
Authors used multiple imputation analysis to have more valid results
Regression Discontinuity Design- Maciejewski, Basu
EDITION ALREADY GRADED A+
UNCERTAINTY AND THE WELFARE ECONOMICS OF MEDICAL CARE- Kenneth Arrow
Research question: How do the conventional principles of welfare economics apply to the healthcare
sector, given the unique characteristics of medical care as an economic good?
Key findings: Healthcare industry differs from standard economic competitive models
information asymmetry: patients have to trust professionals
Market failure: uncertainty, difficulty in assessing quality, moral hazard, demand unpredictable
Role of health insurance: can mitigate inefficiencies and address uncertainty by spreading risk
Public vs private position: need for government intervention
Problems with insurance:administrative costs, predictability, gaps in coverage, trust, licensing
The Economics of Moral Hazard: Comment - Pauly
Goes against Arrow- governemnt creating insurance can lead to inefficiency
People prefer unfair insurance to self insurance- as long as premiums are not too unfair
Prisoner's dilemma leads to moral hazard
deductibles and coinsurance as solution
Even if all individuals are risk averters, some uncertain medical care expenses will not and should not be
insured in an optimal situation
Insurance should exist where demand is inelastic, random , where individuals have risk aversion
Commercial insurance (mandatory) is wrong; everyone has different tastes → different insuranc
Adjustment for Baseline Characteristics in Randomized Clinical Trials- Holmberg, Andersen
Adjusting for baseline (/ stratification) may improve statistical efficiency if they are prognostic and if
stratified randomization is used
should be prespecified, post hoc selection leads to bias
Adjusting can be too complex if too many variables, Stratification can be biased if too many strata, non
prognostic variables can decrease precision
Authors included stratification variable (enrollment site) and 3 pre specified baseline characteristics
, (age, type of admission, organ score)
testing sodium solution on 90 day survival- no effects
Sequential, Multiple Assignment, Randomized Trial Designs Kidwell, Almirall
A SMART is a type of multistage, factorial randomized trial, in which some or all participants are
randomized at 2 or more decision points (treatment A or B), can answer multiple questions
necessary when the optimal sequence of interventions differs among patients
Limitations: no definitive evidence of the effectiveness of an intervention, a SMART is generally not an
adaptive randomized trial, it is common to conflate the variables
The trial: 2-stage, adaptive telecare intervention to treat psychiatric disorders in underserved, rural
settings. Patients randomized to direct vs indirect telecare in stage 1. At the end of stage 1 (month
6),those not engaged were rerandomized in stage 2 to receive a telephone call vs not.
Found no effect for direct telcare, and no effect for phone calls for those not engaged in telcare.
Missing Data: How to Best Account for What Is Not Known- Newgard, Lewis
many methods for handling missing values can yield biased results & decrease study power
complete case analysis, last observation carried forward, baseline observation carried forward, mean
value imputation, random imputation
"informative" when the absence of a value indicates something about what it should be.
MCAR: prob of being missing is unrelated to all observed and unobserved patient characteristics.
MAR: observed values can "explain" which values are missing predict what it would be
MNAR: missing values are dependent on unobserved or unknown factors (most problematic)
Simple imputation methods are "naive" - can introduce bias, and artificially increase precision
limitation of complete case analysis: bias and reduced sample size, reduced study power.
With more missing data, mean value imputation results in greater bias, artificial precision
Random number imputation fails to use observed values to inform the selected estimate.
LOFC assumes that the final outcome does not change from the last observed value.
Authors used multiple imputation analysis to have more valid results
Regression Discontinuity Design- Maciejewski, Basu
