Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
44 Cards in this Set
- Front
- Back
Continuation of Adjusting Costs
|
Continuation of Adjusting Costs
|
|
Patient factors
|
Gender
Age Educational Level Health Insurance Status Medical Condition Culture/ethnicity Lifestyle Personal values and preferences |
|
Non-patient factors
|
Geographic location
Hospital/outpatient practice factors (size type volume of treatments/procedures administered) |
|
Essentially you are leveling the playing field and making sure the outcome
Risk Adjustment Methods: |
Randomization - Before you enroll
Restriction - Before you enroll Stratification - Enroll people in your study and stratify your population male vs female, young vs old and then look at your outcome and see if it differs for those different strata. This is done after the study. Matching - Similar, but you try to enroll someone in group b based on similarity in group a and this is done before you enroll someone in a study. You are deliberately selecting people into groups based on characteristics Scoring - Another post-hoc approach |
|
When you are considering how to risk adjust, you decide based on the data available. You may be limited in type of method based on type of data.
|
Considerations include Availability of data
Timeframe of the data Validity of the data Missing data |
|
Score
|
Data is observational, so you know you need to have to adjust. You level the playing field by scoring their disease burden.
You weight the confounder between treatments a and b. To make sure you are looking at patient a with a similar score to patient b. This tells you whether or not the playing field is level based on disease burden. You can then choose to look at the outcome to see if the outcome differs based on their disease burden score. |
|
Example
|
Most common score you will see is called a charleston co-morbidity index. The goal of the index is to predict likelihood of death.
You have to know that this is an important scoring tool for risk adjusting for mortality risk!!! |
|
Propensity Scoring
|
She skips
|
|
Measuring patient outcomes using surveys - focus on quality of life
|
What do you put in the ratio itself? What outcomes go in? Already talked about cost going in the numerator. Now we will talk about the denominator and what goes in it.
|
|
Focus on quality of life.
Then the next lecture will cover health utility. |
Quality of life is a main outcome used in PE analysis. You can also weight quality of life and weight it for the utility that is gained and these are called UTL's
|
|
To get quality of life you ask the population about their health. To get utility, you ask the population how they value their health. You then weight quality of life with utility.
|
Quality of life is a type of patient reported outcome (PRO).
Patient Reported Outcome - Data tended to be very focused on efficacy (how well does a blood pressure med decrease blood pressure) - very efficacy based trials. Efficacy is good and important, we also need to know how drugs impact people in daily life. FDA struggled with this. What they came up with is that a patient reported outcome is any measure that captures the patients experiences of a disease and its treatments. |
|
This is unique because you are obtained from the patient. They are helpful for diseases where you don't have a good clinical efficacy measure like dementia. A PRO in that setting would be important to obtain their overall quality of life or the caregiver can give their assessment.
|
PRO's are considered a humanistic outcome - any measure relevant to the patient - simply means matters to the human)
|
|
Examples of PROs
|
Health-related quality of life
Functional status Patient satisfaction with treatment Patient preference Work function/employability |
|
Examples of PROs
|
Table - Traditional vs Contemporary trial endpoints
Migraine, Schizo, Alzheimers, Cancer Traditional trial endpoints involve things like efficacy Contemporary trial endpoints involve patient reported outcomes (generally moving in this direction) |
|
Quality of Life is one patient reported outcome, but it is the most popular one.
|
This refers to the persons physical, emotional and social well being - a subjective sense of well-being as opposed to objective circumstances.
|
|
Broad to Specific (look at slide)
|
Humanistic Measures > PROs > QoL > HRQoL
|
|
There is a tremendous growth in quality of life research.
Health related quality of life |
Used to assess the extent to which an illness:
Deteriorates physical and emotional well-being Interferes with social and leisurely activities Disrupts family relationships Causes disability, absenteeism or a decline in work productivity Leads to caregiver burden |
|
Health-related quality of life impacts for dementia may include which of the following:
|
patients reduced physical quality of life
patients reduced mental quality of life caregivers reduced physical quality of life caregivers reduced mental quality of life |
|
When does it make sense to measure quality of life?
|
1) When you have side effects of a treatment that can improve quantity of life, but decrease quality of life (chemotherapy)
2) When treatment improves patients ability to function in home, work or leisurely activities 3) Social/emotional suffering may be as bad or worse than physical suffering - If you were doing a traditional clinical trial, you may focus on pain. In some cases, its not a simple of physical suffering but also social and emotional suffering. Depression and pain are common co-morbidity's |
|
Examples of when quality of life endpoints can be better indicators of health than clinical measures
|
GERD - patients often report symptoms of heartburn and reflux despite negative test results.
Anemia in End Stage Renal Disease (ESRD) - quality of life may be low even when anemia measures (Hgb/Hct) are within target ranges |
|
How do we measure quality of life?
|
Health Profile Measures
-SF-36 (Very popular, common clinical trials) -Nottingham Health Profile -Sickness Impact Profile Health Utility or Preference Measures - Discussed in health utility lecture |
|
Hierarchy of Quality of Life Measurement
|
Items are grouped into Domains which are then summarized in a Summary score
Items ask about how the patient can physically function. They then fall in the physical domain. Social questions fall into the social relationships domain etc. Domains then grouped into the summary score which represents overall quality of life. |
|
True or False - Domains generally include 1 question.
|
False - Group of questions based on the concept that they cover
|
|
SF-36 - Most popular quality of life tool. 36 Items
8 Domains = Physical Functioning, Role-Physical, Bodily Pain, General Health Vitality, Social Functioning, Role-Emotional, Mental Health |
Two scores
One represents how treatment impacts physical health One represents how treatment impacts mental health Some surveys combine everything into one score. Not the SF-36. You then take these scores and use them in your PE ratios. |
|
Would not test on the number of items per domain.
Important to know that this is a very popular quality of life tool, captures physical social and emotional well being has a two main scores and can be used in a PE analysis or one of its shorter versions. |
Skipped Sickness Impact Profile
Only need to know that its a quality of life tool like SF-36, but is much longer - 136 items Not used very often. |
|
Nottingham Health Profile
|
Same general idea
Not as popular as SF36 Grouped domains is a big different, but the same quality of life constructs are the same Just know that this is a quality of life tool. Nothing specific. |
|
These tools are considered a general quality of life tool because they are not particular to a disease. Your finding how Health impacts quality of life, not a disease. Sometimes you want to do more than a general tool will give you.
|
Then you want to use a Disease-Specific Health Related Quality of Life Instrument.
Example: The SF-36 would ask have you been able to climb the stairs and what what extent? Disease specific qol question would ask for a COPD patient, does your COPD impact your ability to climb the stairs? The items mention the disease. |
|
Not very much literature supporting the use of Disease-specific tools.
|
Once you've committed yourself to one disease, you can't compare to another disease - limiting characteristics.
|
|
Desirable Attributes of Health Related Quality of Life Instruments
|
What should you be looking for in the instrument to assess that it measures what its supposed to.
|
|
Key things to look for
|
Validity
Reliability Responsiveness to change These are referred to as psychometric properties |
|
Validity
|
The accuracy of the measure in reflecting the concept its supposed to measure
"The truth, what you want to be measuring, the correct answer you are seeking" |
|
Reliability
|
How reproducible results are
"Administer instrument to same person and get the same score over and over" |
|
Responsiveness
|
"Step on a scale, then eat a lot and step on it again and it shows that you gained weight - that is responsiveness - the scale responds to a change in weight"
|
|
Bulls-eye diagrams of Reliability vs Validity
|
High validity/High reliability - cluster of dots in the middle (Best)
Low validity/High reliablity - cluster of dots away from middle (3rd best) - Results are at least consistent, but not close to the truth. Moderate validity/Low reliability (2nd best) - Not perfectly reliable but some validity Neither a strong validity nor reliability (worst) |
|
Types of Validity - measuring the truth
Face Content Construct Concurrent At a minimum, you want to do face validity, but ideally several of them |
Face - instrument appears to the a good measure at face value (simply looking at the instrument and judging it measures what you believe it should measure - normally experts in the field will be asked to inspect it)
|
|
Content - How representative your instrument is of the entire concept
Quality of life - covers social, emotional and physical well being If you came up with a qol instrument that came up with questions about their social and physical but not emotional function, it would have low content validity |
How well the instrument behaves as you expect it to behave
|
|
Construct - concepts that are within the instrument
Seek to determine whether or not those concepts change between different groups of people or circumstances. qol - social emotional physical - if administered to people who are sick, score to be low, healthy - scores are high |
That instrument would have a good construct validity - its behaving as you predict it should. The constructs (concepts) are responding the way they should
Can an anxiety instrument pick up the level of symptoms or just presence or absence. You would field the tool to two groups with known level differences and see if there is a different severity between the groups. If you can differentiate between levels, it has good construct validity. |
|
Concurrent - You are looking at how well the instrument corresponds to a measure of the same concept. You are looking at the correlation between the correlation of your instrument to the gold standard
|
Difficult to measure in the absence of a gold standard.
|
|
Types of Reliablity - reproducibility
Major types = Test - Retest, Internal consistency |
Test-Retest = scores are stable upon administering the test more than once.
You need to make sure that your population has not changed. You would never want to do a test-retest with a cancer group over a long period of time. It is a aimed to be a short term assessment where you administer the test several times over a short time period. If you know your population is stable and consistent in health, you could then do test-retest over a long period of time. |
|
Sometimes this needs to be done outside of a clinical trial, because people are getting one treatment or another or placebo and their health will change over time, unless you have an ideal situation where you can re-administer the test several times over a short period of time.
|
Internal consistency - How well the items hinge together within the instrument itself. How reliable the items are in the instrument.
Looking at the concepts that you are measuring (physical, emotional, social) and questions for each concept. You want to make sure these types of questions are correlated to each other. Example: Measuring physical functioning. Difficulty getting up the stairs, Difficulty dressing self. You would expect the person who can't get up the stairs to also say they can't dress themselves. You want to makes sure the way people answer makes sense. There wouldn't be internal consistency if the person could get up the stairs but not dress themselves. |
|
Measuring responsiveness - You want your patient populations health to change to see if the test is responsive.
|
This is critical for an instrument designed to evaluate treatment effects or diseases where symptoms wax and wane.
|
|
Questions to ask when considering whether to use an existing survey instrument:
|
- Is the tool general or disease-specific? (General - you can compare health status of people with different diseases)
- Is the language appropriate for my study population? (In britain they call a bladder infection a water-works infection) - Is the response burden acceptable? (If the response burden is more than 30 min, it is not a desirable tool to use. Peoples attention wanes with a long questionnaire) |
|
- What concepts are captures? (Looking at the different concepts and making sure the concepts are relevant for the topic - emotional and social but not physical functioning)
- Has it been validated in a population similar to my study? (Even if there is validity data, is the data related to your population. Example: You have data from a medicaid population and you are interested is the tool valid in a private payer population) - How strong is the supporting reliability (One data supporting the tool is not enough, you would want 2, 3, 4 or more to say it is valid and reliable) |
Defining the clinically meaningful difference in instrument scores- One of the biggest challenges is that they dont know what a meaningful difference between scores is.
A 5 point improvement - depending on what perspective you may or may not agree this is important for everyone. They struggle with defining what is clinically meaningful. --> Clinicians perspective - leads the clinician to change their therapy --> Patients perspective - change in score that prompts them to seek healthcare or request a change Statistically significant doesn't mean its going to be meaningful to the population. |
|
Not going to be tested on last 2 slides.
|
DONE
|