Title: Patient Reported Outcomes aka: Health Related Quality of Life
1Patient Reported Outcomesaka Health Related
Quality of Life
- Lowell E. Schnipper, M.D.
- Berenson Professor of Medicine
- Harvard Medical School
- Chief, Hematology-Oncology
- Beth Israel Deaconess Medical Center
2Targeted Therapy
- EGFR pathway
- When the tumor isnt the only target (so whats
new?)
3Patient Reported OutcomesWhy?
- Cancer often an incurable illness
- Improvement in patients HRQoL is an expectation
of appropriate therapy - Palliation is often the goal
- Can be a vital help in comparative effectiveness
analyses - HRQoL is the product of Patient Reported Outcomes
4 Flow of symptom information in cancer treatment
trials
Trotti, A. et al. J Clin Oncol 255121-5127 2007
5 Adverse effects domains
Trotti, A. et al. J Clin Oncol 255121-5127 2007
6Value to the Clinical Trials Process
- Provide information about toxicity
- Symptom relief
- These often are primary or secondary clinical
trial endpoints - In the clinical setting HRQoL assessments can
have predictive value for survival - Useful for comparing various therapies for a
simlilar indication
7Subjectivity and Objectivity
- HRQoL is not subjective in the usual sense of the
term - It can be measured accurately in an individual,
and in a group - It is subjective in that it derives from the
human subject of research or clinical practice
8Health-Related Quality of Life (HRQOL) Assessment
1. What is HRQOL? 2. What Should You Measure? 3.
How Should You Measure it? 4. How is it
incorporated into clinical research
9P. Ganz, personal communicationt
10HRQoL Domains
- Physical Functioning Spirituality
- Occupational Role
- Functioning Future Orientation
- Social Functioning Sexuality/Intimacy
- Emotional Well-being
Health Concerns - Symptom Status Family Well-Being
- Financial Concerns Satisfaction with care
- Global/Overall Perception of Quality of Life
11Types of QOL Instruments
- Health profiles--descriptive questionnaires-measur
e by a simple metric different aspects of HRQoL
across multiple domains - Generic, multi-dimensional
- Disease-specific, multi-dimensional
- Condition-specific, multi-dimensional
-
12Definitions
- Instrument Questionnaire
- Item Question
- Domain Dimension Area of Behavior or
Experience - Cross-Sectional DesignCompares 2 or More
Populations at One Point in Time - Longitudinal DesignAssesses Changes in HRQOL
Over Time in 1 or More Populations
13HRQOL Assessment
What Should You Measure?
14(No Transcript)
15Functional Status
- Focus May be Generic or Disease-Specific
- usually in multiple domains
- Physical
- Emotional
- Social
16Emotional State
Linear Analog Self-Assessment
Place an X at a point on the line that best
represents how you have felt over the past two
weeks.
X
17Aggregated Scores
Simple Average
Sum or Mean of Item or Subscale Scores to Yield a
Global Score This Approach Weights each Item
Proportional to its Variance in the Population
Under Study Example Quality of Life Index
18Quality of Life Index
19Quality of Life Index
20HRQOL
Functional Status
Symptoms
Global Ratings
21Symptoms
- Focus
- Disease (e.g., asthma, CHF, cancer, lung cancer)
- Type of symptom (e.g., pain)
- Dimension(s)
- Frequency
- Severity
- Distress
- Interference with activities
22HRQOL
Functional Status
Symptoms
Global Ratings
23Global Ratings
Single Assessment of Overall Health or HRQOL
Example E-V-G-F-P
How would you describe your overall state of
health CHECK ONE? Excellent Very Good
Good Fair Poor
24HRQOL Assessment
Why Should You Measure It?
25HRQOL Value Added
Operational Definition
HRQOL Assessment Adds Value IF HRQOL Data Can
Influence Overall Conclusions of the Study
26Study Goals
Nature of the Value Added Depends on Overall
Study Goals
- evaluating the effectiveness of interventions
- characterizing treatment-specific outcomes for
use in shared decision making - characterizing the burden of illness
- predicting patient outcomes
- quality-adjusting survival for resource
allocation and other policy decisions
27Question
- HRQoL is relevant to which type(s) of clinical
trials - Phase I
- Phase Ib/II
- Randomized phase II trial
- Prospective randomized trial
- All of the above
28Evaluating Effectiveness
Depends On
1. Phase of the Trial 2. Natural History of
Disease
29Type of Clinical Trial and PRO
- Phase I not critical-goal is primarily to define
MTD or optimal biologic dose - Phase II can be used but not essential
- An opportunity to collect pilot data for use in
phase III trial - Phase III-HRQOL data is essential in this context
-
30Tailoring study design to the natural history of
the disease
- 3 paradigms
- Chronic and/or recurrent, non-lethal diseases
- Lethal but curable diseases
- Lethal, incurable diseases
31Supporting SharedDecision Making
Information Patients Need
Nature of Alternative TreatmentsLikely Outcomes
from Treatments Descriptive Information forEach
of the Possible Outcomes simple, transparent
measures
32Predicting Outcomes
- Baseline HRQoL has been Shown to be an
Independent Predictor for -
- overall survival - overview of 36 trials that
assessed baseline PROs and mortality (Gotay, JCO
261355, 2009) - PRO is a complex biomarker that can be highly
predictive must be better incorporated into the
design and analysis of comparative trials
33HRQoL Assessment
How Should You Measure It?
Identify the Respondent patientsurrogateprovider
(MD, RN, other)
34HRQoL Assessment
How Should You Measure It?
Choose the Domains to be Measured disease or
domain-specific instrumentgeneric
instrumentglobal assessment of QoL
35HRQoL Assessment
How Should You Measure It?
- Identify the Respondent
- Choose the Domains to be Measured
Consider the Need for Interpretability health
state descriptionstranslating numeric into
clinical differences
36HRQoL Assessment
How Should You Measure It?
- Identify the Respondent
- Choose the Domains to be Measured
- Consider Need for Interpretability
Choose a Mode of Administration in person, by
phone, mailedcomputer assisted (in person or
phone)self-administered (forms, computer)
37HRQoL Instruments
- Multi-dimensional HRQoL instruments are available
that cover the basic HRQoL domains - Generic
- Cancer-specific
- Domain-specific
38HRQoL Instruments
- Generic types
- Medical Outcome Study Short Form Health Survey
(SF-36) Ware, JE, Med Care (1992)30473-83. - EuroQoL (EQ-5D) Health Policy (1990)16199-208.
- Intended for use across broad chronic disease
populations - Allow comparisons across these groups
- Disadvantage may not permit adequate
cancer-specific focus - Disease caused symptoms
- Treated related symptoms
39HRQoL Instruments
- Cancer-specific instruments
- Responsive to disease-related changes
- Cannot be used across populations with chronic
disease - Several are in common use because they are
reliable and have been validated - Functional Living Index-Cancer ShipperJCO
(1984) 2427-483 - Modular instruments combine a generic or core
instrument applicable to a broad range of cancer
patients with cancer-type specific questions - FACT-C Gunnars, Acta Oncologica (2001)40175-84.
- EORTC QLQ-C30, Aaronson JNCI (1993)85365-367.
40HRQoL Instruments
- Domain Specific Instruments
- Designed to address one specific aspect of HRQoL
- Examples
- Multi-dimensional Fatigue Inventory (MFI)
- McGill Pain Questionnaire
41HRQOL Assessment
Other Considerations
- Is the proposed analysis too taxing for the
subject? - must questionnaire be completed serially in light
of possible attrition - Is a proxy needed?
- will population include children, low literacy,
non-English speaking, visually impaired, or
cognitively impaired adults?
42HRQOL Assessment
Other Considerations (contd)
- timing of assessments
- consistency across patients/groups critical
- responses influenced by recent experiences
- missing data
- avoid at all costs
- generally cannot be assumed to be at random
- specify methods for dealing in advance!
- multiple comparisons
- specify primary endpoints in advance
43HRQOL Assessment
Make Your Final Choice
- Psychometric (responsive, valid, reproducible)
properties - validated in your population
- known and used in your field
44Patient Reported Outcomes Methodological
Challenges
- The variety of questionnaires available
- Choice of time points
- Incomplete data (data attrition)
- Lack of pre-defined endpoints
- Response-shift over time of patient perceptions
of HRQoL - Psychological defenses tend to conserve
perception of good HRQoL
45Question
- In a longitudinal study the best way to deal
with response shift is to - Increase the sample size to dilute this effect
- Anticipate it by changing the scales used at
longitudinal time points when making HRQoL
measurements - Use a pre-test (then test) and post-test set of
questions at a follow-up visit to assess the
degree of change (shift) in baseline attitude of
HRQoL - None of the above
46Examples from Completed Trials
47Developing a HRQoL instrument PrinciplesPhase
III Breast Ca Trial
- 1. Dimensions Physical, emotional and social
well-being - 2. Importance items (questions) must reflect
importance to patients - 3. Quantifiable summary scores must be amenable
to statistical analysis - 4. Validity the instrument should be a true
measurement of HRQoL - 5. Reproducibility the instrument should yield
similar results in comparable patients - 6. Responsiveness the instrument should detect
clinically important changes - 7. Simplicity the questionnaire should be short
- Shipper, JCO 61798-1810, 1988
48Developing a HRQoL instrument Principles
- Item selection
- Interview patients and determine what issues are
important - Winnow redundancies
- Breast cancer Questionnaire (BCQ) hair loss,
emotional dysfunction, physical symptoms, trouble
and inconvenience associated with treatment,
fatigue, nausea, positive well-being - BCQ contained 30 questions,
- Likert 7 point scale administered by an
interviewer, and took 10-15 minutes patients
asked to recall how they had been feeling over a
defined period (2 weeks in this case). Maximum
score on scale equates with most minimal deficit - Final score expressed as mean of the score on the
30 questions - Shipper, JCO 61798-1810, 1988
49Developing a HRQoL instrument Principles
- Co-administration of several additional
questionnaires, e.g., Spitzer QoL, Rand Physical
and Emotional Status, Karnofsky score (by
physician) - Each questionnaire was administered at intervals
throughout study treatment period and up to 72
weeks. - Shipper, JCO 61798-1810, 1988
50Developing a HRQoL instrument Principles
- BCQ had the best correlations with the Rand
Physical and Emotional surveys and the Spitzer
QoL instrument (Correlation coefficients approx.
0.6) - Not surprisingly the prolonged treatment
demonstrated a significant dip in QoL using the
BCQ and Karnofsky scores during the extended
treatment period. The BCQ correlated better with
the patients assessments of her QoL. - Shipper, JCO 61798-1810, 1988
51Prednisone /- Mitoxantrone in Hormone Resistant
Prostate Cancer
- Randomized controlled trial
- Palliation as an end point 2 point reduction in
the McGill-Melazak Pain questionnaire - Approval of Mitoxantrone for this indication
based upon positive QoL findings
Tannock, JCO (1996)141756-64
52Prednisone /- Mitoxantrone in Hormone Resistant
Prostate Cancer
- Instruments employed
- Prostate Cancer-Specific QoL Instrument with 9
linear self-assessment scales (pain, physical
activity, appetite, constipation, etc.) - EORTC/QLQ-C30 contains multi-item domains for
physical function, emotional function, pain,
global QoL
Tannock, JCO (1996)141756-64
53Prednisone /- Mitoxantrone in Hormone Resistant
Prostate Cancer
- Results
- Favored MP for pain (38 reduction), physical
activity, constipation and mood - QoL parameters (PS, pain intensity and serum alk
Phos) significant predictors of OS---PSA was not! - Could be expanded to include cost effectiveness
(cost per quality adjusted life year gained)
Tannock, JCO (1996)141756-64
54Question
- In the mitoxantrone/Prednisone trial just
discussed, how would do you account for missing
data?
55Question
- a. if there is extensive dropping out, increase
the sample size - b. the sample size should be large enough to
dilute the effect of missing data - c. anticipate loss of patient availability and
identify proxies at the start of the study - d. use proxies at the outset, rather than the
patient, due to the anticipated high rate of
patients dropping out
56Quality-Adjusted Survival
- QALYs Preferred Outcome for CEA
- Quality of Life Must be Measured inSuch a Way
that Product of Length of Life and Quality of
Life is Meaningful - UTILITIES
- e.g., 1 year of life at quality x must be as
desirable as 6 months of life at quality 2x
57Utilities
Quantitative Measure of Strength of a Persons
Preference for an Outcome
- measured on a scale of 0-1
- 0 represents death worst conceivable state
- 1 represents perfect health
- key difference reflects how a subject values a
health state, not just the characteristics of
that health state
58Quality-Adjusted Survival
1
Utility
0
Time
Area under the curve quality-adjusted survival
59Utilities
Rating Scale
- visual analog scale, feeling thermometer,or
verbal rating scale - e.g., If 0 is death and 100 is perfect health,
what number would you say best describes your
current state of health? - NOT a true utility
60Perfect Health
__________ 100 __________ __________ 99
__________ __________ 98 __________ __________
97 __________ __________ 96
__________ __________ 95 __________ __________
94 __________ __________ 93
__________ __________ 92 __________ __________
91 __________ __________ 90
__________ __________ 89 __________ __________
88 __________ __________ 87
__________ __________ 86 __________ __________
85 __________ __________ 84
__________ __________ 83 __________ __________
82 __________ __________ 81
__________ __________ 80 __________ __________
79 __________ __________ 78
__________ __________ 77 __________ __________
76 __________ __________ 75
__________ __________ 74 __________ __________
73 __________ __________ 72
__________ __________ 71 __________ __________
70 __________ __________ 69
__________ __________ 68 __________ __________
67 __________ __________ 66
__________ __________ 65 __________ __________
64 __________ __________ 63
__________ __________ 62 __________ __________
61 __________ __________ 60
__________ __________ 59 __________ __________
58 __________ __________ 57
__________ __________ 56 __________ __________
55 __________ __________ 54
__________ __________ 53 __________ __________
52 __________ __________ 51
__________ __________ 50 __________ __________
49 __________ __________ 48
__________ __________ 47 __________ __________
46 __________ __________ 45
__________ __________ 44 __________ __________
43 __________ __________ 42
__________ __________ 41 __________ __________
40 __________ __________ 39
__________ __________ 38 __________ __________
37 __________ __________ 36
__________ __________ 35 __________ __________
34 __________ __________ 33
__________ __________ 32 __________ __________
31 __________ __________ 30
__________ __________ 29 __________ __________
28 __________ __________ 27
__________ __________ 26 __________ __________
25 __________ __________ 24
__________ __________ 23 __________ __________
22 __________ __________ 21
__________ __________ 20 __________ __________
19 __________ __________ 18
__________ __________ 17 __________ __________
16 __________ __________ 15
__________ __________ 14 __________ __________
13 __________ __________ 12
__________ __________ 11 __________ __________
10 __________ __________ 9
__________ __________ 18 __________ __________
7 __________ __________ 6
__________ __________ 5 __________ __________
4 __________ __________ 3
__________ __________ 2 __________ __________
1 __________ __________ 0 __________
Death
61Utilities
Standard (reference) Gamble (SG)
- assesses the utility of a health state by asking
how much one would risk to improve it - respondent asked to choose between life in a
given health state and a gamble between death and
perfect health - utility is given by the
probability of perfect health in the gamble when
the respondent is indifferent between the gamble
and life in the given health state - affected by attitude towards risk
62Utilities
Time Tradeoff (TTO)
- assesses the utility of a health state by asking
how much time one would give up to improve it - respondent is offered a choice between a set
length of life in a given compromised health
state and a shorter length of life in perfect
health - utility is given by the ratio of life
expectancy in perfect health to life expectancy
in the given health state when the respondent is
indifferent
63Utilities
Practical Considerations
- Utility Questions are Challenging
- hypothetical situations
- involves probabilities
- awful outcomes
- SG and TTO Cannot be Self-Administered
- Best Done Using Iterative Approach
- Visual Aids Can Be Helpful
64Health Indexes
HRQOL Questionnaire
Descriptive Data
Patient
Formula
Utility
Reference Population
Pickard, Value in Health, 2009
65CONCLUSIONS
1. Integration of QOL outcomes in cancer clinical
trials is challenging 2. Anticipation of design,
data collection, and analysis will decrease some
problems 3. Additional resources are required to
incorporate QOL outcomes successfully-budget for
them!
66(No Transcript)