Date post: | 28-Dec-2015 |
Category: |
Documents |
Upload: | samuel-harrell |
View: | 264 times |
Download: | 7 times |
INTERGRATION OF QUALITY OF LIFE AND SURVIVAL FOR COMPARATIVE HEALTH
RISK/OUTCOME ASSESSMENT
Jung-Der Wang, M.D., Sc. D.
National Taiwan University College of Public Health
National Taiwan University Hospital
OUTLINES• Estimating health expectancy in Taiwan
• Revisiting the integration of survival and quality of life (QOL), and concept of QALY (quality-adjusted life year) as a common unit for risk/outcome evaluation and cost-effectiveness, especially quantifying health benefits of prevention of a disease by QALY
• Extending to psychometric measurement for QOL and clinical decision making
• Integrating with medical cost of the NHI
Problems in direct comparison between EU and Taiwan for HLY:Taiwan: (National Health Survey) surveyed age: 12-85; self-administered questionnaire ; SF-36 and WHOQOL-BREF (age 12-65)
EU countries: Are you restricted in daily activities as a result of
longstanding illness, condition or handicap? All the time (severe); now and then (moderate);
seldom or no (no disability)
SF-36 physical function• The following items are about activities
you might do during a typical day. Does your health now limit you in these activities? If so, how much?
a. (Vigorous activities, such as running, etc.)
b. Moderate activities, such as moving a table, etc.
c to j. Lifting or carrying groceries; climbing several flights of stairs; climbing one flight of stairs, bending or stooping; walking > one block; bathing or dressing yourself.
Yes, limited a lot; Yes, limited a little; No, not limited at all
SF-36 physical function Yes, limited a lot; Yes, limited a little;
No, not limited at all (male)
0
10
20
30
40
50
60
70
80
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85
total life expectancydisability-freewith disability
0102030405060708090
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86
total life expectancydisability-freewith disability
SF-36 physical function Yes, limited a lot; Yes, limited a little;
No, not limited at all (female)
SF-36 physical function Yes, limited a lot; Yes, limited a little; No, not limited at all
0
10
20
30
40
50
60
70 bad
fair
good
0246810121416182022
Taiwan
SF-36 physical function (male) Yes, limited a lot; Yes, limited a little; No, not limited at all
0246810121416182022
Taiwan
SF-36 physical function (female) Yes, limited a lot; Yes, limited a little; No, not limited at all
Redefining health care (2006): by Michael Porter and Elizabeth Teisberg
• Value is the health outcomes per dollar spent in providing services. Outcomes are multidimensional, and include not only survival but quality of life, extent of recovery or disability, errors, complications, recovery time, recurrences, and other aspects of the patient’s health experience.
Evidence based medicine:
• There is no room for spending money on ineffective diagnosis and treatment for any medical condition.
• Preventive medicine must also be evidence-based
• Quantification of health benefit of preventing the occurrence of a specific disease
(How much loss of health benefit
for a specific disease ?)
No. articles in PubMed database with two specific key words
0
2000
4000
6000
8000
10000
12000
14000
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Quality of Life Evidence based medicine
394 7
6621
13053
0
2000
4000
6000
8000
10000
12000
14000
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Quality of Life Evidence based medicine
394 7
6621
13053
0
2000
4000
6000
8000
10000
12000
14000
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Quality of Life Evidence based medicine
394 7
6621
13053
A common question raised:• Is there a common unit to measure
both the survival and utility or psychometry of quality of life?
• Live vs. Dead ---- counting the no. of lives saved
• More delicate measures:
--Length of survival S(t) or S(ti|xi) --Quality of life Qol(ti|xi)
• Can we measure S(ti|xi) or Qol(ti|xi)?
• Can we develop a method to combine both?
• (Can we quantify the cost paid by the NHI? )
Estimated survival function, mean QOL and quality adjusted survival curve; The area under the QAS curve is the expected quality adjusted survival time (Hwang JS, et al Statistics in Medicine 1996;15:93-102)
)dt()]([ tStQolEQALE Quality-Adjusted life expectancy (QALE)
Health-Adjusted life expectancy (HALE)
Qol(t): quality of life function at time t
k different health states hk with weight Wk at time t
dttShWHALE k,tk,t
k)(][
k
kk hW ,
Can be calculated by the mean of quality of life at time t (in terms of utility)
wk: weight of health state k
hk: utiluty value of health state k
life ofquality of valueexpected )]([ tQolE
male
0
0.2
0.4
0.6
0.8
1
12 17 22 27 32 37 42 47 52 57 62 67 72 77age
survivalQOL
female
0
0.2
0.4
0.6
0.8
1
12 17 22 27 32 37 42 47 52 57 62 67 72 77 age
survivalQOL
Notation of a typical life table with added columns of QOL (quality of life) and QAST (quality adjusted survival time)
. . . . . . . . .
. . . . . . . . .
. . . . . . . .
. . . . . . . . .
1 0
ConditionalProportionSurviving
CumulativeProportionSurvivingInterval
NumberLost toFollow-up
NumberWithdrawn Alive
NumberDying
NumberEnteringInterval
NumberExposedTo Risk
ConditionalProportion Dying
QOL ti( ) QAST
21 tt l1 w1 d1 n'1 n1 1q̂ 1p̂ 00.1)(ˆ 1 ts qol t( )1 QS1
32 tt l 2 w2 d 2 n'2 n2 2q̂ 2p̂ )(ˆ 2ts qol t( )2 QS 2
1 ii tt li wi di n i' ni iq̂ ip̂ )(ˆ its qol ti( ) QSi
ss tt 1 ls 1 ws 1 ds 1 n s' 1 ns 1 1ˆ sq 1ˆ sp )(ˆ 1sts qol ts( ) 1 QSs 1
st ls ws ds n s' ns )(ˆ sts qol ts( ) QSs
A more general model:
• xi:determinant(s) of S(survival) and U(utility) functions e.g. head injury, stroke,….., etc.
• Quality adjusted survival Qol(t| xi): quality of life function
(Wang JD. Basic principles and practical applications in epidemiological research. 2002)
)dt|()]|([ ii xtSxtUE
)dt|()]|([ ii xtSxtQolE
Cost of illness approach:
• Human capital left over for determinant xi
WA(t| xi): work ability function
• Direct medical cost of determinant xi
Cost(t| xi): medical cost function
)dt|()]|([ ii xtSxtWE A
)dt|()]|([ ii xtSxtCostE
Gain of health benefits from successful prevention of a disease xi
• Let x0 denotes age- and gender- matched referents simulated from the general population or vital statistics
•
__
)dt|()]|([ ii xtSxtQolE
)dt|()]|([ 00 xtSxtQolE
ILLUSTRATIVE EXAMPLES:
• How much utility of health (in QALY) does it cost for a case of end stage renal disease or liver cancer?
• --- Survival curve• --- Quality of life estimation--- General population of Taiwan in
1995 as the reference population assuming QOL=1
Utility (SG) for Utility measures of HCC
Time in Months
Qu
alit
y a
dju
ste
d s
urv
iva
l
0 10 20 30 40 50 60
0.0
0.2
0.4
0.6
0.8
1.0
Utility measured by standard gamble (SG)
Time in Months
QA
S
0 100 200 300 400 500 600
0.0
0.2
0.4
0.6
0.8
1.0
Reference populationHCC cohort
Shaded area =233.6 QALM loss due to liver cancer
Pit dug for washing underground soil and water
RISK:LIKELIHOOD OF EVENT
(Incidence rate or probability)
X
CONSEQUENCE OF EVENT
(loss of utility due to the event)
(need to establish a cohort to estimate)
Cancer risks based on RME (reasonable maximal
exposure) and cancer slopes
Vinyl chloride QALM
8.4 x 10-6 (X 233.6 .002
Tetrachloroethylene QALM) =
1.9 x 10-4 .044
Trichloroethylene
1.4 x 10-4 .032
IF there are 1000 people at risk, then the above numbers must be multiplied with 1,000
Example 2: helmet law for motor cycle riders
Utility gained from the enforcement of helmet law
• Number of cases of head injury prevented by wearing a helmet
--- full face vs. half- face covered• Number of QALY lost per case of head
injury --- case registry of head injury in
Taipei city followed for 7 years(Tsauo JY, et al. Accident Analysis & Prevention 1999;31:253-63)
The survival function of the head injury cases and the reference population in the 80 months after onset
0
0.5
1
1 9 17 25 33 41 49 57 65 73
months of follow-up
surv
ival
fu
nct
ion
referencepopulationcasepopulation
The health-related quality of life of the head injury cases and the reference population in the 80 months after onset
0
0.2
0.4
0.6
0.8
1
1.2
1 9 17 25 33 41 49 57 65 73
months of follow-up
he
alt
h-r
ela
ted
qu
alit
y o
f lif
e
referencepopulationcasepopulation
0
0.2
0.4
0.6
0.8
1
1.2
1 8 15 22 29 36 43 50 57 64 71 78
months of follow-up
qu
alit
y a
dju
sted
su
rviv
al t
ime
referencepopulationcasepopulation
The quality-adjusted survival time of the head injury cases and the reference population in the 80 months after onset
EXTENSION TO HEALTH PROFILE (PSYCHOMETRIC
SCORE)• Consequence of the event
can be replaced by QOL measured by psychometrics
• Hwang JS, Wang JD. Quality of Life Research 2004; 13:1-10
Psychometric mean score
• The sum of scores of those who are still alive plus those who die
• The following simple equation establishes the relationship between population mean QoL score function and survival function,
where Qs(t) is the average QOL of surviving subjects at time t
)](1[)()()( tStQtStQ s
Estimations
• The estimate of expected psychometric score-adjusted survival (PAS) for an index population,
is obtained by firstly estimatingand at chosen time points ’s
maxmax
00d))(1(d)()(][
TT
s ttSttQtSPASE
)(tS)(tQs
kt
Survival-weighted Health Profile in Long-term Survivors of Acute Myelogenous Leukemia (AML)
Chiun Hsu1, Jung-Der Wang1, Jing-Shiang Hwang2, and Jih-Luh Tang 1
National Taiwan University Hospital1
Academia Sinica2
Taipei, Taiwan
(Quality of Life Research 2003 ;12:503-517)
Comparison of life time psychometric Scores for BMT and chemotherapy (WHOQOL generic instrument)
Comparison of life time psychometric scores for BMT and chemotherapy(WHOQOL generic measurement)
HOW MUCH DOES IT COST FOR A UNIT OF SCORE-TIME?
• Through questioning 157 patients with disability caused by occupational injury under contingent valuation method or stated preference, we found that people are willing to pay US$ 65.1-69.6 for a pain-killer pill that can remove pain for 24 hours.
Ho JJ, et al. et al. (monetary value of score time) Accident Analysis & Prevention 2005;37:537-48
• The WTP money for removing a longer duration of pain is even bigger
Lifetime cost (NT$) for hemodialysisAnnual cost
Discount rate
Lifetime cost
mean median mean median
Out patient clinic
476,553 606,800 0% 3,870,084 4,927,820
2% 3,303,139 4,205,923
4% 2,890,398 3,680,375
Hospitalization 43,133 24,600 0% 350,279 199,776
2% 298,965 170,510
4% 261,608 149,204Total 422,863 578,100 0% 3,434,073 4,694,748
2% 2,931,001 4,006,994
4% 2,564,760 3,506,303
Conclusion: for outcome/risk assessment in health and medicine
• The QALY or life year gained or loss plus the psychometric score time can be estimated for comparative assessment of health risks/outcomes in national health resources allocation and clinical decision makings (and for cost-effectiveness analysis).
• Measurements of QOL had better be improved to an interval scale.
• Life-time utility (Economist)
經濟學家:終生預期效用
survival function人命 ( 存活函數 ) utility function --HRQL(健康相關生活品質 ) --working ability, wages, medical costs 工作能力、薪資、醫療費用
• Quality-adjusted life expectancy or healthy life expectancy (生活品質調整後預期壽命 )
)dt|()]|([ xitSxitUE
)dt|()]|([ xitSxitQolE
Clinical decision making
Maximize individual patient’s utility under resource constraint
based on: psychometric theory
WHOQOL health profile (multi-dimensions)
+
survival function
National resource allocation
Maximize utility of all people (No. of QALY) under the constraint of National Insurance
System (NIS)
based on: expected utility theory
EQ-5D (or other utility measurement)
+
survival function
survival weighted psychometric scores for each facets
maxmax
00d))(1(d)()(][
TT
s ttSttQtSPASE QAS (QALY)= )|())|(( ii xtSxtQolE
Each patient participates in clinical decision to maximize no. of QALY/per given cost
How much is the patient willing to pay?
Cost / QALY(or DALY)
How much will NIS pay per QALY under the constraint of distributive justice?
Summarize to only one dimension
Hwang JS, Tsauo JY, Wang JD. (theory of QAS) Stat Med 1996;15:93-102
Hwang JS, Wang JD. (QAS extrapolation to lifetime) Stat Med 1999;18:1627-40
Tsauo JY, et al. (Utility of enforcement of helmet law) Accident Anal Prev 1999;31:253-63
Yao KP, et al. (WHOQOL-BREF Taiwan version) J Formos Med Assoc 2002;101:342-51
Lee LJH, et al. (Risk assessment for water pollution) J Toxicol Environ Health 2002;65:219-35
Hwang JS, Wang JD (extended to psychometry) Quality Life Res 2004; 13:1-10
Hsu J, et al. (bone marrow transplantation for leukemia) Qual Life Res 2003 ;12:503-517
Chuang HY, et al. (occupational health policy for lead) J Toxicol Environ Health 2005; 68:1485-96.
Ho JJ, et al. (monetary value of score time) Accident Anal Prev 2005;37:537-48.
Ho JJ, et al. (survival of occupational disability) Scand J Work Environ Health 2006; 32(2):91-98.
Ho WL, et al. (survival and cost of thalassemia) Bone Marrow Transplant 2006; 37(6):569-574.
Ho JJ, et al. Estimation of reduced life expectancy. Accident Anal Prev 2006; 38:961-968.
Fang CT et al. (Life expectancy of patients with HIV/AIDS). Quarterly J Med 2007; 100:97-105.
Fang CT et al. (Cost-effectiveness for HAART policy) J Formos Med Assoc 2007; 106(8):631–640
Chu PC et al. (Lifetime financial burden to the National Health Insurance for 17 different cancer in Taiwan) J Formos Med Assoc 2008; 107:54-63
Chu PC et al. (Life expectancy and loss of life expectancy for major cancer in Taiwan) Value in Health 2008; 7:1102-1109
Chang CY et al (Quality of life in obese patients) Obesity Surg 2008; DOI 10.1007/s11695-008-9513-z
THANK YOU FOR YOUR ATTENTION
Extrapolation of survival under high censored rate: Semi-parametric modeling (Hwang & Wang 1999, Fang et al. 2007)
H (t | patient) = H (t | reference) + constant excess hazard C1
)population reference|()populationpatient |()( tStStW
logit W(t) = ln [exp (C0 – C1 × t)/(1 – exp (C0 – C1 × t))]
= C0 – C1 × t – ln [1 – exp (C0 – C1 × t)]Because C1 > 0, the residual item ln [1 – exp (C0 – C1 × t)] will converge to 0 when t . As a result, when t , logit W(t) will approximate to C0 – C1 × t , which is a straight line with a slope of – C1.
Total Hazard Background Hazard
age- and gender-matched
0 20 40 60
Time in Months
02
46
810
Logi
t W(t)
0 20 40 60
Time in Months
24
68
10Lo
git W
(t)
0 20 40 60
Times in Months
0.0
0.2
0.4
0.6
0.8
1.0
Pro
babi
lity
of S
urvi
val
Actually observedSemi-parametricGamma modelWeibull model
0 20 40 60
Times in Months
0.0
0.2
0.4
0.6
0.8
1.0
Pro
babi
lity
of S
urvi
val
Actually observedSemi-parametricGamma modelWeibull model
AIDS group
Non-AIDS group
3-year survival extrapolated to 6 years
3-year survival extrapolated to 6 years
Cancer site Cohort size
Mean age at Dx (SD)
Life expectancy
Years of life loss
Lifetime cost (3% discount)
Pancreas 7,931 65.6 (12.7) 2.81 (0.17) 12.87 263,000
Lung 58,773 66.6 (11.7) 3.09 (0.07) 11.79 342,000
Liver 68,585 60.4 (13.5) 3.45 (0.08) 15.61 238,000
Esophagus 9,710 63.0 (12.1) 3.54 (0.20) 13.25330,000
Gallbladder & bile duct
5,097 66.5 (12.0) 4.98 (0.20) 10.36446,000
Stomach 35,477 64.9 (13.6) 7.51 (0.14) 8.80 609,000
Prostate 14,288 73.1 (8.0) 8.17 (0.13) 1.72 527,000
Oral cavity 26,681 53.8 (12.9) 9.58 (0.61) 14.00910,000
Cancer site Cohort size
Mean age at Dx (SD)
Life expectancy
Years of life
loss
Lifetime cost (3% discount)
Colon & rectum
60,789 63.8 (13.7) 10.86 (0.11) 6.36 692,000
Kidney & urinary tract
11,671 62.7 (15.1) 10.97 (0.85) 6.74 528,000
Bladder 15,092 66.7 (12.6) 10.99 (0.20) 3.83 519,000
Leukemia 9,224 41.8 (25.5) 11.61 (0.94) 19.34 2,404,000
Nasopharynx 15,231 49.6 (13.4) 12.59 (0.74) 14.79 632,000
Skin 14,005 63.3 (16.9) 16.16 (0.22) 1.59 354,000
Ovary 6,436 49.3 (17.0) 17.71 (0.80) 11.91 1,277,000
Cervix uteri 29,636 54.7 (13.8) 19.77 (0.30) 6.18 808,000
Breast 36,668 50.5 (12.5) 20.01 (0.80) 9.35 1,081,000
Domain Facets Epilepsy(yes/no)
Frequency of seizure
Marriage Co-morbid
Environ-ment
Health and social care
0.32** (0.06)
Financial resources
0.22* (0.08)
−0.017** (0.006)
Participation in recreation
−0.014* (0.006)
−0.30* (0.12)
Opportunities for new skills
−0.017** (0.005)
Physical safety and security
−0.013* (0.006)
−0.28* (0.11)
Social Personal relationships
−0.30** (0.07)
−0.014* (0.005)
0.21** (0.06)
−0.23* (0.10)
Being respected −0.27** (0.06)
−0.010* (0.005)
0.23** (0.06)
Sexual activity −0.24** (0.07)
0.47** (0.08)
* p < 0.05** p < 0.005
Quality of life of epilepsy patients (Liu HH, et al. Epilepsy Res 2005)
Domain Facets BMI(25–32)
BMI(32–35)
BMI(35–40)
BMI(>40)
Employ-ment
Physical Pain and discomfort
−0.33*(0.14)
Energy and fatigue
−0.37*(0.19)
−0.57**(0.18)
−0.56**(0.11)
Sleep and rest −0.51*(0.22)
−0.51**(0.20)
−0.64**(0.12)
Psychol-ogical
Thinking & concentration
−0.60**(0.21)
−0.69**(0.20)
−0.53**(0.12)
0.43**(0.11)
Self-esteem −0.59**(0.20)
−0.54**(0.18)
−0.84**(0.11)
0.32**(0.10)
Body image & appearance
−1.13**(0.21)
−1.32**(0.20)
−1.35**(0.12)
Sexual activity −0.47*(0.18)
−0.43**(0.17)
−0.54**(0.10)
0.24**(0.09)
Being respected
−0.52**(0.18)
−0.41**(0.11)
0.36**(0.10)
* p < 0.05** p < 0.005
Quality of life in obese patients (Chang CY et al. Obesity Surg 2008)
Transformation of Scale Score• Transform each raw scale score to a 0 to 100 scale using the
formula shown below.
• Formulas for scoring and transforming scales
• ExampleA physical functioning raw score of 21 would be transformed as follows: [(21-10)/20]*100=55Where lowest possible score=10 and possible raw score range =20
100
rangescorerawpossible
scorerawpossiblelowestscorerawActualScaledTransforme
Scale Sum Final Item Values (after recoding items)
Lowest and highest possible raw scores
Possible raw score range
Physical functioning
3a+3b+3c+3d+3e+3f+3g+3h+3i+3j
10,30 20