SLIDE 1
Co nstruc ting I nve rse Pro b a b ility We ig hts fo r Sta tic I - - PowerPoint PPT Presentation
Co nstruc ting I nve rse Pro b a b ility We ig hts fo r Sta tic I - - PowerPoint PPT Presentation
Co nstruc ting I nve rse Pro b a b ility We ig hts fo r Sta tic I nte rve ntio ns K unja l Pa te l, DSc MPH Se nio r Re se a rc h Sc ie ntist Ha rva rd T .H. Cha n Sc ho o l o f Pub lic He a lth Ac kno wle dg e me nt Slide s c o
SLIDE 2
SLIDE 3
Ca se Study
SLIDE 4
I ntro duc tio n/ Ba c kg ro und
Ra ndo mize d c o ntro lle d tria ls (RCT s) in HI V-infe c te d a dults ha ve sho wn HAART to b e hig hly e ffe c tive in re duc ing the risk o f mo rta lity
Na tura l pro g re ssio n o f HI V infe c tio n in c hildre n is diffe re nt fro m a dults:
HI V RNA le ve ls re ma in pe rsiste ntly hig he r tha n a dults fo r first 2-3 ye a rs o f life , de c re a sing to ste a dy sta te le ve ls in a dults a fte r a ppro xima te ly five -ye a rs
Ge ne ra liza b ility o f a dult tria l re sults?
SLIDE 5
Studie s o f HAART in Childre n
RCT s o f HAART in c hildre n ha ve fo c use d o n inte rme dia te immuno lo g ic a nd viro lo g ic e ndpo ints
L
- ng -te rm studie s o f HAART
- n mo rta lity re lia nt o n
- b se rva tio na l studie s:
I ta lia n c o ho rt study: triple c o mb ina tio n the ra py vs. no the ra py – HR=0.29 (0.13-0.67) (De Martino e t al. 2000)
PACT G 219 study: c o mb ina tio n the ra py with PI vs. the ra py witho ut PI – HR=0.33 (0.19-0.58) (Go rtmake r e t al. 2000)
SLIDE 6
Ne e d to Re -E va lua te E ffe c t o f HAART
- n Mo rta lity
Pre vio us studie s e nde d fo llo w-up in 1999:
Use o f ne w a ntire tro vira l drug s ha s inc re a se d
Cha ng e s in initia l HAART re g ime ns o ve r time
Va n Dyke e t a l. JAIDS 2011; 57:165-173
SLIDE 7
Study q ue stio n
Wha t is the e ffe c t o f HAART
- n mo rtality a mo ng
pe rina ta lly HI V-infe c te d c hildre n?
I
s this a g o o d study q ue stio n?
SLIDE 8
F
- rmula tio n o f a we ll-de fine d study q ue stio n
We ll-de fine d c a usa l infe re nc e q ue stio ns c a n b e ma ppe d
into a ta rg e t tria l
Ca se e xa mple : Wha t is the e ffe c t o f initiating HAART
- n
mo rta lity a mo ng pe rina ta lly HI V-infe c te d c hildre n?
Spe c ify the pro to c o l o f the ta rg e t tria l inc luding : E
lig ib ility c rite ria
T
re a tme nt stra te g ie s
Ra ndo mize d tre a tme nt a ssig nme nt F
- llo w-up pe rio d
Outc o me Ca usa l c o ntra st o f inte re st Ana lysis Pla n
He rna n, Ro b ins Am J E pid e mio l. 2016;183(8):758–764
SLIDE 9
Pe dia tric AI DS Clinic a l T ria ls Gro up (PACT G) Pro to c o ls 219 & 219C
Pro spe c tive c o ho rt studie s o f HI
V-e xpo se d c hildre n (infe c te d a nd uninfe c te d) fro m mo re tha n 80 study site s in the US
Asse ss the lo ng -te rm e ffe c ts o f HI
V infe c tio n a nd in- ute ro a nd po stna ta l e xpo sure to a ntire tro vira l the ra py
PACT
G 219: April 1993-Se pte mb e r 2000
PACT
G 219C: Se pte mb e r 2000-2006
E
xte nsive c linic a l, ne uro psyc ho lo g ic a l, a nd la b o ra to ry e va lua tio ns
SLIDE 10
Study Po pula tio n, E xpo sure , F
- llo w-up
1,236 pe rina ta lly HI
V-infe c te d c hildre n e nro lle d in PACT G 219 a nd 219C b e twe e n Ja nua ry 1, 1996 a nd June 30, 2006
E
xc lude s tho se with pre vio us o r c urre nt use o f HAART a t time o f study e ntry
HAART
de fine d a s the use o f a t le a st 3 drug s fro m a t le a st 2 diffe re nt c la sse s o f HI V the ra py (NRT I s, NNRT I s, o r PI s)
Onc e sub je c ts initia te d HAART
the y we re a ssume d to re ma in o n HAART fo r the dura tio n o f the ir fo llo w-up
F
- llo w-up fo r a ma ximum o f te n ye a rs to the la st visit a t
whic h sub je c t wa s se e n a live o r the la st visit b e fo re June 30, 2006 (i.e . “c o mple tio n o f study”)
SLIDE 11
Cla ssific a tio n o f tre a tme nt stra te g ie s a c c o rding to the ir time c o urse
Point inte rve ntio ns
I
nte rve ntio n o c c urs a t a sing le time
E
xa mple s: o ne -do se va c c ina tio n, sho rt-live d tra uma tic e ve nt, surg e ry…
I
nte ntio n-to -tre a t e ffe c ts in RCT s a re a b o ut po int inte rve ntio ns
Sustaine d stra te g ie s
I
nte rve ntio ns o c c ur a t se ve ra l time s
E
xa mple s: me dic a l tre a tme nts, life style , e nviro nme nta l e xpo sure s…
Ma ny (mo st? ) q ue stio ns a re a b o ut susta ine d e xpo sure s
SLIDE 12
Cla ssific a tio n o f susta ine d tre a tme nt stra te g ie s
Static a fixe d str
ate gy for e ve r yone
E
xample : tr e at with 150mg of daily aspir in dur ing 5 ye ar s
Case e xample : initiate HAART Dyna mic a stra te g y tha t a ssig ns diffe re nt va lue s to diffe re nt
individua ls a s a func tio n o f the ir e vo lving c ha ra c te ristic s
E
xa mple : sta rt a spirin tre a tme nt if c o ro na ry he a rt dise a se , sto p if stro ke
Ca se e xa mple : initia te HAART
if CD4 dro ps b e lo w 500 c e lls/ mm3
SLIDE 13
Ra ndo mize d tre a tme nt a ssig nme nt
Ca usa l infe re nc e me tho ds a re me tho ds tha t e mula te
ra ndo miza tio n
Why is ra ndo miza tio n impo rta nt?
SLIDE 14
De finitio n o f a n a ve ra g e c a usa l e ffe c t
l
E
a c h pe rso n ha s two c o unte rfa c tua l o utc o me s:
Outc o me Y if tre a te d - Yi, a=1 Outc o me Y if untre a te d – Yi, a=0
I
ndividua l c a usa l e ffe c t:
Yi, a=1 ≠ Yi, a=0 Ca nno t b e de te rmine d e xc e pt unde r e xtre me ly stro ng
a ssumptio ns
Ave ra g e (po pula tio n) c a usa l e ffe c t:
E
[Ya=1 = 1] ≠ E [Ya=0 = 1]
Ca n b e e stima te d unde r: No a ssumptio ns (ide a l ra ndo mize d e xpe rime nts) Stro ng a ssumptio ns (o b se rva tio na l studie s)
SLIDE 15
Ca usa tio n ve rsus Asso c ia tio n
Pr[Ya=0=1] Pr[Ya=1=1] Pr[Y=1|A=0] Pr[Y=1|A=1]
l
SLIDE 16
Ca usa tio n ve rsus Asso c ia tio n
Pr[Ya=1]
pro po rtio n o f sub je c ts tha t wo uld ha ve de ve lo pe d the
- utc o me Y ha d a ll sub je c ts in the po pula tio n re c e ive d
e xpo sure va lue a
(Co unte rfa c tua l) risk o f Ya Unc o nditio na l o f ma rg ina l pro b a b ility – “c a lc ula te d” using
da ta fro m the who le po pula tio n
Ca usa tio n: Pr[Ya=1=1] ≠ Pr[Ya=0 = 1]
Pr[Y=1|A=a]
Pro po rtio n o f sub je c ts tha t de ve lo pe d o utc o me Y a mo ng
tho se tha t re c e ive d e xpo sure va lue a in the po pula tio n
Risk o f Y a mo ng tho se e xpo se d/ une xpo se d Co nditio na l pro b a b ility – c a lc ula te d b y using da ta fro m a
sub se t o f the po pula tio n
Asso c ia tio n: Pr[Y=1|A=1] ≠ Pr[Y=1|A=0]
SLIDE 17
I de a l Ra ndo mize d E xpe rime nt
L
a rg e (ne a r-infinite ) po pula tio n
No lo ss to fo llo w-up F
ull c o mplia nc e (a dhe re nc e ) to a ssig ne d e xpo sure o r tre a tme nt
Do ub le b lind a ssig nme nt
SLIDE 18
Ra ndo miza tio n (I )
Assume two e xpo sure g ro ups (tre a te d a nd untre a te d) Me mb e rship in e a c h g ro up is ra ndo mly a ssig ne d
e .g ., b y a flip o f a c o in
F
irst o ptio n:
T
re a t sub je c ts in g ro up 1, do n’ t tre a t sub je c ts in g ro up 2
Pr[Y=1|A=1] is, sa y, 0.57
Se c o nd o ptio n:
T
re a t sub je c ts in g ro up 2, do n’ t tre a t sub je c ts in g ro up 1
Wha t is the risk? Pr[Y=1|A=1] is ?
0.57
SLIDE 19
Ra ndo miza tio n (I I )
Whe n g ro up me mb e rship is ra ndo mly a ssig ne d, risks a re
the sa me
Bo th g ro ups a re c o mpa ra b le o r e xc hange able E
xc ha ng e a b ility is the c o nse q ue nc e o f ra ndo miza tio n
SLIDE 20
E xc ha ng e a b ility
Sub je c ts in g ro up 1 wo uld ha ve ha d the sa me risk a s
tho se in g ro up 2 ha d the y re c e ive d the tre a tme nt o f tho se in g ro up 2
T
he c o unte rfa c tua l risk a mo ng the tre a te d e q ua ls the c o unte rfa c tua l risk a mo ng the untre a te d unde r the sa me e xpo sure le ve l
Pr[Ya=1|A=1] = Pr[Ya=1|A=0] A Ya Ya A I
mplie s la c k o f c o nfo unding
SLIDE 21
I n ide a l ra ndo mize d e xpe rime nts
Pr[Y=1|A=1] is e q ua l to Pr[Ya=1=1] Pr[Y=1|A=0] is e q ua l to Pr[Ya=0=1]
T
he re fo re the a sso c ia tio na l risk ra tio
Pr[Y=1|A=1]/ Pr[Y=1|A=0]
is e q ua l to the c a usa l risk ra tio
Pr[Ya=1=1]/ Pr[Ya=0=1]
SLIDE 22
Why is Pr[Y=1/ A=1] is e q ua l to Pr[Ya=1=1]
A two ste p pro o f:
1. Pr[Y=1|A=1] = Pr[Ya=1=1|A=1]
b y de finitio n o f a c o unte rfa c tua l va ria b le (i.e ., c o nsiste nc y)
2. Pr[Ya=1=1|A=1] = Pr[Ya=1=1|A=0] = Pr[Ya=1=1]
b y ra ndo miza tio n – (i.e ., e xc ha ng e a b ility)
Ste p 2 no t g e ne ra lly true in the a b se nc e o f ra ndo miza tio n
SLIDE 23
I n a n ide a l ra ndo mize d e xpe rime nt
Asso c ia tio n is c a usa tio n
Be c a use ra ndo miza tio n pro duc e d e xc ha ng e a b ility
We ha ve a me tho d fo r c a usa l infe re nc e !
No ne e d fo r a djustme nts o f a ny so rt Assumptio n-fre e !
….Ho we ve r, re a l ra ndo mize d e xpe rime nts a re no t ide a l ra ndo mize d e xpe rime nts….No c le a r-c ut se pa ra tio n b e twe e n re a l ra ndo mize d e xpe rime nts a nd
- b se rva tio na l studie s…
SLIDE 24
De a d e nd?
E
xc ha ng e a b ility (a c o nse q ue nc e o f ra ndo miza tio n) is a c o nditio n fo r c a usa l infe re nc e
E
xc ha ng e a b ility is no t g e ne ra lly a n a c c e pta b le a ssumptio n in o b se rva tio na l studie s
E
xpo se d a nd une xpo se d g e ne ra lly no t c o mpa ra b le
I
ndividua ls who re c e ive a he a rt tra nspla nt ma y ha ve mo re se ve re dise a se
Ca se e xa mple : c hildre n who initia te HAART
ma y ha ve mo re se ve re dise a se tha n tho se who do n’ t (i.e ., c o nfo unding b y indic a tio n)
A c o nditio n we a ke r tha n e xc ha ng e a b ility is ne e de d fo r
c a usa l infe re nc e fro m o b se rva tio na l da ta
SLIDE 25
Ho pe
Co nside r o nly individua ls with the sa me pre -e xpo sure
pro g no stic fa c to rs
T
he n the e xpo se d a nd une xpo se d ma y b e e xc ha ng e a b le
e .g ., a mo ng individua ls with a n e je c tio n fra c tio n o f 10%,
tho se who do a nd do no t re c e ive a he a rt tra nspla nt ma y b e c o mpa ra b le
e .g ., a mo ng individua ls with CD4 c o unt <100, tho se who do
a nd do no t re c e ive a ntire tro vira l the ra py ma y b e c o mpa ra b le
T
his is o fte n re a so na b le
E
spe c ia lly if c o nditio ning o n ma ny pre -e xpo sure c o va ria te s L
SLIDE 26
Co nditio na l E xc ha ng e a b ility
Within le ve ls o f the c o va ria te s, L
, e xpo se d sub je c ts wo uld ha ve ha d the sa me risk a s une xpo se d sub je c ts ha d the y b e e n une xpo se d, a nd vic e ve rsa
Co unte rfa c tua l risk is the sa me in the e xpo se d a nd the
une xpo se d with the sa me le ve l o f L
Pr[Ya=1|A=1, L
=l] = Pr[Ya=1|A=0, L =l] A Ya|L =l Ya A|L =l
E
q uiva le nt to ra ndo miza tio n within le ve ls o f L
I
mplie s no unme a sure d (re sidua l) c o nfo unding within le ve ls o f the me a sure d c o va ria te s L
SLIDE 27
I n a n o b se rva tio na l study
Asso c ia tio n is c a usa tio n within le ve ls o f the c o va ria te s
Unde r the assumption o f c o nditio na l e xc ha ng e a b ility
We ha ve a me tho d fo r c a usa l infe re nc e fro m
- b se rva tio na l da ta tha t is no t a ssumptio n-fre e
But the ne e d to re ly o n this a ssumptio n is no t T
HE pro b le m
SLIDE 28
T HE pro b le m
T
he a ssumptio n o f c o nditio na l e xc ha ng e a b ility is
unte stable
E
ve n if the re is c o nditio na l e xc ha ng e a b ility, the re is no wa y we c a n kno w it with c e rta inty
T
his is why c a usa l infe re nc e fro m o b se rva tio na l da ta is c o ntro ve rsia l
We c a n use e xpe rt kno wle dg e to e nha nc e pla usib ility o f the
a ssumptio n
Me a sure a s ma ny re le va nt pre -e xpo sure c o va ria te s a s
po ssib le
Ca n o nly ho pe tha t the a ssumptio n is a ppro xima te ly true
(i.e ., the re ma y b e c o nfo unding due to unme a sure d fa c to rs)
SLIDE 29
Me tho ds to c o mpute c a usa l e ffe c ts
Stra tific a tio n Re g re ssio n Ma tc hing Sta nda rdiza tio n I
nve rse pro b a b ility we ig hting AL L a ssuming c o nditio na l e xc ha ng e a b ility
SLIDE 30
Cho ic e o f me tho d de pe nds o n type o f stra te g ie s
Co mpa riso n o f stra te g ie s invo lving po int inte rve ntio ns
- nly
All me tho ds wo rk if a ll b a se line c o nfo unde rs a re me a sure d Co mpa riso n o f susta ine d stra te g ie s Ge ne ra lly o nly c a usa l infe re nc e me tho ds wo rk T
ime -va rying tre a tme nts imply time -va rying c o nfo unde rs
po ssib le tre a tme nt-c o nfo unde r fe e db a c k Co nve ntio na l me tho ds ma y intro duc e b ia s e ve n whe n
suffic ie nt da ta a re a va ila b le o n time -va rying tre a tme nts a nd time -va rying c o nfo unde rs
SLIDE 31
Ca se e xa mple : HAART initia tio n o ve r time
SLIDE 32
Ca se E xa mple : Dire c te d Ac yc lic Gra ph
Whe re L i = c o nfo und e r (CD4, vira l lo a d , e tc ) info rma tio n a t time i, Ai = tre a tme nt (HAART ) info rma tio n a t time i, a nd Y = o utc o me (mo rta lity) info rma tio n a t time i. U = unme a sure d c o va ria te
L0 A0 L1 A1 Y U
SLIDE 33
Pro b le m with Stra tifie d Ana lytic Appro a c h
L A0 L
1
A1 Y1
U
Inte re ste d in the c umula tive e ffe c t o f tre a tme nt. L 1 is a c o nfo unde r fo r the tre a tme nt A1 – if do n’ t a djust fo r it the n tre a tme nt
e ffe c t is c o nfo unde d.
Also c o uld induc e se le c tio n b ia s (c o llide r). L 1 is a ffe c te d b y A0 – if a djust fo r L 1 the n lo sing so me o f the e ffe c t o f A0.
SLIDE 34
I nve rse pro b a b ility we ig hting
YOU will c o mpute the c a usa l risk ra tio using
inve rse pro b a b ility we ig hting (I PW) in a n
- b se rva tio na l study
i.e ., yo u will c o mpute Pr[Ya=1=1]/ Pr[Ya=0=1]
unde r c o nditio na l e xc ha ng e a b ility
SLIDE 35
A simplifie d o b se rva tio na l study
500 HI
V-infe c te d a dults
Va ria b le s:
L
=1: CD4 c e ll c o unt <200 c e lls/ mm 3
A=1: o n hig hly a c tive a ntire tro vira l the ra py (HAART
)
Y=1: AI
DS
T
re a tme nt sta tus is de c ide d a fte r lo o king a t CD4 c e ll c o unt
No lo ss to fo llo w-up
SLIDE 36
T he da ta summa rize d in a ta b le
L =0 L =1 Y=1 Y=0 Y=1 Y=0 A=1 15 35 144 216 A=0 30 20 32 8
SLIDE 37
T he da ta summa rize d in a tre e
32 30 35 15 8 20 216 144 50
L =1: CD4 c e ll c o unt <200 c e lls/ mm3
A=1: o n HAART
Y=1: AI DS
SLIDE 38
Yo ur g o a l
T
- c o mpute the e ffe c t o f HAART
- n the risk o f AI
DS o n the c a usa l risk ra tio sc a le
Pr[Ya=1=1]/ Pr[Ya=0=1] Assuming c o nditio na l e xc ha ng e a b ility within le ve ls o f L
F
irst, c o mpute Pr[Ya=0=1]
Se c o nd, c o mpute Pr[Ya=1=1]
SLIDE 39
Orig ina l da ta
32 30 35 15 8 20 216 144 50
L =1: CD4 c e ll c o unt <200 c e lls/ mm3
A=1: o n HAART
Y=1: AI DS
SLIDE 40
Da ta ha d e ve ryo ne b e e n untre a te d
320 60 80 40
L =1: CD4 c e ll c o unt <200 c e lls/ mm3
A=1: o n HAART
Y=1: AI DS
SLIDE 41
Da ta ha d e ve ryo ne b e e n tre a te d
70 30 240 160 100
L =1: CD4 c e ll c o unt <200 c e lls/ mm3
A=1: o n HAART
Y=1: AI DS
SLIDE 42
320 60 70 30 80 40 240 160 100
W=1/f[A|L]
1/.5=2 1/.5=2 1/.5=2 1/.5=2 1/.1=10 1/.1=10 1/.9=1.11 1/.9=1.11
Da ta ha d e ve ryo ne b e e n tre a te d a nd untre a te d
Pseudopopulation
SLIDE 43
Pse udo po pula tio n da ta a na lysis
Ya=1 Ya=0 a =1 a =0 190 380 310 120
Pr[Ya=1=1] = 190/ (190+310) = 0.38 Pr[Ya=0=1] = 380/ (380+120) = 0.76 Ca usa l risk ra tio = 0.38/ 0.76 = 0.5 YOU DI
D I T ! Co mpute d the c a usa l risk ra tio using I PW
SLIDE 44
Whic h a ssumptio n a re yo u ma king ?
Ya A|L =l
Co nditio na l e xc ha ng e a b ility in the po pula tio n
E
xpo sure is ra ndo mize d within le ve ls o f L
No unme a sure d c o nfo unding within le ve ls o f the me a sure d
va ria b le L
Within le ve ls o f L
, the risk a mo ng the e xpo se d if the y we re une xpo se d is the sa me a s the risk a mo ng the une xpo se d in the po pula tio n
a nd vic e ve rsa
SLIDE 45
Unde r c o nditio na l e xc ha ng e a b ility
T
he o b se rva tio na l study in the o rig ina l po pula tio n is a ra ndo mize d e xpe rime nt within le ve ls o f L
T
he study in the pse udo po pula tio n c re a te d b y I PW is a ra ndo mize d e xpe rime nt
E
xpo se d a nd une xpo se d sub je c ts a re (unc o nditio na lly) e xc ha ng e a b le b e c a use the y a re the sa me individua ls
E
xpo sure is ra ndo mize d (i.e . e q ua lly pro b a b le a c ro ss le ve ls o f the c o va ria te L )
T
he re is no c o nfo unding
I
n the pse udo po pula tio n, c a usa l e ffe c ts c a n b e e stima te d a s in a ra ndo mize d e xpe rime nt
No ne e d fo r a djustme nt o f a ny so rt
SLIDE 46
Dire c te d Ac yc lic Gra ph in Pse udo po pula tio n
L0 A0 L1 A1 Y U
SLIDE 47
Use o f mo de ls fo r I PW
Re a lity is we de a l with hig h-dime nsio na l da ta with
multiple c o va ria te s (L s), so me with multiple le ve ls
Ca nno t o b ta in me a ning ful no n-pa ra me tric e stima te s o f the
we ig hts
Mo de l the pro b a b ility o f e xpo sure with L
s a s the c o va ria te s
So me individua ls ma y c o ntrib ute a re a lly hig h we ig ht due
to the ir a re la tive ly sma ll pro b a b ility o f ha ving the e xpo sure the y ha d g ive n the ir c o va ria te histo ry
Sta b ilize the we ig hts b y using the pro b a b ility o f tre a tme nt in
the nume ra to r
Apply sta b ilize d we ig hts (SW) to e stima te the pa ra me te rs o f
a ma rg ina l struc tura l mo de l
re duc e va ria nc e in mo de l fo r the o utc o me
SLIDE 48
Sta b ilize d I nve rse Pro b a b ility o f T re a tme nt We ig hts
Nume ra to r: T
he pro b a b ility tha t the sub je c t re c e ive d his/ he r o b se rve d tre a tme nt a t we e k k, c o nditio na l o n pa st tre a tme nt histo ry a nd b a se line c o va ria te s.
De no mina to r: T
he pro b a b ility tha t the sub je c t re c e ive d his/ he r o wn o b se rve d tre a tme nt a t we e k k, g ive n pa st tre a tme nt histo ry a nd c o va ria te histo ry (b a se line a nd time -de pe nde nt).
SLIDE 49
Dire c te d Ac yc lic Gra ph in Pse udo po pula tio n with SW
V A0 L1 A1 Y U
SLIDE 50
E stima ting I PW a nd fitting the MSM
E
stima te SW fo r b o th tre a tme nt a nd c e nso ring :
F
it lo g istic re g re ssio n mo de ls fo r tre a tme nt a nd c e nso ring
Use pre dic te d va lue s fro m the mo de ls to c a lc ula te sta b ilize d
we ig hts
E
stima te the I PW e stima te o f HAART
- n mo rta lity:
F
it we ig hte d po o le d lo g istic mo de l using the e stima te d sta b ilize d we ig hts.
Use “ro b ust” va ria nc e e stima to rs (GE
E ) to a llo w fo r c o rre la te d o b se rva tio ns c re a te d b y we ig hting – c o nse rva tive 95% CI .
SLIDE 51
Ca se E xa mple : Co nfo unde rs
- Ag e
- Se x
- Ra c e / E
thnic ity
- We e k o f F
- llo w-up
- Ca le nda r Ye a r
- CDC Clinic a l Ca te g o ry
- CD4%
- T
- ta l lympho c yte c o unt
- White b lo o d c e ll c o unt
- He ma to c rit
- Alb umin
SLIDE 52
SAS Co de fo r E stima ting Nume ra to r a nd De no mina to r fo r T re a tme nt I PW
SLIDE 53
SAS Co de fo r E stima ting Nume ra to r a nd De no mina to r fo r Ce nso ring I PW
SLIDE 54
SAS Co de fo r Ca lc ula ting I PW a nd sta b ilize d I PW (1)
SLIDE 55
E xa mple Da ta
k1_0 phrt_0 k1_w phrt_w k2_0 punc_0 k2_w punc_w IPTWnum TrtPredictNum IPTWden TrtPredictDen IPCWnum CensPredictNum IPCWden CensPredictDen 49360 0.96375 0.96375 0.9658 0.9658 0.98868 0.98868 0.97466 0.97466 49360 13 0.92911 0.96406 0.93651 0.96967 0.97643 0.98761 0.95619 0.98105 49360 26 0.89599 0.96436 0.90843 0.97002 0.96318 0.98643 0.93807 0.98105 49360 39 0.86433 0.96466 0.8815 0.97036 0.94888 0.98515 0.9203 0.98105 49360 52 0.83404 0.96496 0.85273 0.96735 0.93346 0.98375 0.89698 0.97466 49360 65 1 0.02898 0.96525 0.03368 0.96051 0.92481 0.99074 0.87097 0.97101 49360 78 1 0.02898 . 0.03368 . 0.91543 0.98986 0.84129 0.96592 49360 91 1 0.02898 . 0.03368 . 0.90526 0.98889 0.81261 0.96592 49360 104 1 0.02898 . 0.03368 . 0.89426 0.98784 0.77494 0.95364 49360 117 1 0.02898 . 0.03368 . 0.88235 0.98669 0.73902 0.95364 49360 130 1 0.02898 . 0.03368 . 0.8695 0.98543 0.71683 0.96997 49360 143 1 0.02898 . 0.03368 . 0.85564 0.98406 0.6836 0.95364 49360 156 1 0.02898 . 0.03368 . 0.84071 0.98255 0.65191 0.95364 49360 169 1 0.02898 . 0.03368 . 0.82466 0.98091 0.62169 0.95364 49360 182 1 0.02898 . 0.03368 . 0.80743 0.97912 0.59287 0.95364 49360 195 1 0.02898 . 0.03368 . 0.78899 0.97716 0.56539 0.95364 49360 208 1 0.02898 . 0.03368 . 0.76928 0.97502 0.53918 0.95364 49360 221 1 0.02898 . 0.03368 . 0.74827 0.97269 0.48733 0.90384 49360 234 1 0.02898 . 0.03368 . 0.72593 0.97015 0.45571 0.93512 49360 247 1 0.02898 . 0.03368 . 0.70225 0.96737 0.43398 0.95232 49360 260 1 0.02898 . 0.03368 . 0.67722 0.96436 0.41329 0.95232 49360 273 1 0.02898 . 0.03368 . 0.65085 0.96107 0.3863 0.93471 49360 286 1 0.02898 . 0.03368 . 0.62318 0.95749 0.35013 0.90637 49360 299 1 0.02898 . 0.03368 . 0.59427 0.9536 0.32502 0.92826 49360 312 1 0.02898 . 0.03368 . 0.56418 0.94937 0.3017 0.92826 49360 325 1 0.02898 . 0.03368 . 0.53303 0.94478 0.28005 0.92826 49360 338 1 1 0.02898 . 0.03368 . 0.50095 0.9398 0.26063 0.93063 Patid Week HAART Censored
SLIDE 56
E xa mple da ta wo rkshe e t
k1_0 phrt_0 k1_w phrt_w k2_0 punc_0 k2_w punc_w IPTWnum TrtPredictNum IPTWden TrtPredictDen IPCWnum CensPredictNum IPCWden CensPredictDen 49488 0.96972 0.96972 0.97553 0.97553 0.98239 0.98239 0.94649 0.94649 49488 13 0.9406 0.96998 0.9557 0.97967 0.96347 0.98074 0.9094 0.96081 49488 26 1 0.028 0.97023 0.02285 0.97609 0.95287 0.989 0.85239 0.93731 49488 39 1 0.028 . 0.02285 . 0.9414 0.98796 0.79402 0.93153 49488 52 1 0.028 . 0.02285 . 0.92899 0.98682 0.74591 0.93941 49488 65 1 0.028 . 0.02285 . 0.91559 0.98557 0.69915 0.93731 49488 78 1 0.028 . 0.02285 . 0.90113 0.98421 0.65532 0.93731 49488 91 1 1 0.028 . 0.02285 . 0.88556 0.98272 0.6145 0.93771 Patid Week HAART Censored
SLIDE 57
E xa mple da ta wo rkshe e t – c a lc ula te SW
k1_0 k1_w k2_0 k2_w Stabilized IPTWnum IPTWden IPCWnum IPCWden Weight 49488 0.96972 0.97553 0.98239 0.94649 1.03175 49488 13 0.9406 0.9557 0.96347 0.9094 1.04272 49488 26 0.028 0.02285 0.95287 0.85239 1.37007 49488 39 0.028 0.02285 0.9414 0.79402 1.45308 49488 52 0.028 0.02285 0.92899 0.74591 1.52641 49488 65 0.028 0.02285 0.91559 0.69915 1.605 49488 78 0.028 0.02285 0.90113 0.65532 1.68531 49488 91 0.028 0.02285 0.88556 0.6145 1.76621 Patid Week
SLIDE 58
SAS Co de fo r F ina l MSM
SLIDE 59
E stima te d E ffe c t o f HAART
- n Mo rta lity fro m
Unwe ig hte d (Sta nda rd) a nd We ig hte d Mo de ls
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Unweighted model, no covariates Unweighted model, baseline covariates Weighted model, baseline covariates Unweighted model, baseline and time-dependent covariates Hazard ratio (95% CI)
SLIDE 60
Assumptio ns fo r I PW e stima tio n o f a MSM
Co nditio na l e xc ha ng e a b ility within le ve ls o f me a sure d
c o va ria te s
Una b le to a djust fo r HI
V-1 vira l lo a d
Use d to g uide de c isio ns a b o ut whe n to initia te HAART
in re c e nt ye a rs a nd is a sso c ia te d with mo rta lity
Re po rte d HR like ly to b e unde re stima te d
Co rre c t mo de l spe c ific a tio n fo r a ll mo de ls to e stima te
we ig hts a nd fina l MSM
SLIDE 61
Co nc lusio ns
L
- ng -te rm HAART
use (> 5 ye a rs) is a sso c ia te d with sig nific a ntly lo we r mo rta lity a mo ng c hildre n a nd a do le sc e nts infe c te d with HI V-1 c o mpa re d to no n-HAART use .
Suppo rt c urre nt US pe dia tric g uide line s Re sults c o mpa ra b le to a dult RCT
a nd pre vio us pe dia tric
- b se rva tio na l studie s
Suppo rt e xpa nde d de live ry o f c a re to HI
V-infe c te d c hildre n g lo b a lly
SLIDE 62
Co nc lusio ns
Co ntinue d fo llo w-up is ne e de d a s this po pula tio n a g e s
a nd ma ture s with the use o f HAART
Ne e d to e stima te the e ffe c ts o f pro lo ng e d use o f HAART
- n
immune func tio n, g ro wth, se xua l ma tura tio n, a nd q ua lity o f life pa ra me te rs
SLIDE 63