increasing value, saving lives: health care in a new era - keynote address by brent james
DESCRIPTION
Keynote address by Brent James at the 2013 Saskatchewan Health Care Quality Summit. For more information about the summit, visit www.qualitysummit.ca. Follow @QualitySummit on Twitter. Dr. Brent James describes how Intermountain Healthcare is systematically, and successfully, bringing together clinicians, patients and leaders to: establish best practices; drive out waste in their system; and ultimately deliver better, safer care. Dr. James will share insights about the structures, strategies and relationships that have been pivotal in transforming their health system.TRANSCRIPT
Increasing Value, Saving Lives:Health Care in a New Era
Brent C. James, M.D., M.Stat.Executive Director, Institute for Health Care Delivery ResearchIntermountain HealthcareSalt Lake City, Utah, USA
Saskatchewan Health Quality CouncilSaskatchewan Health Care Quality Summit 2013
Evraz Place, Regina, Saskatchewan, CanadaThursday, 11 April 2013 -- 8:15a - 9:45a
Disclosures
Neither I, Brent C. James, nor any family members, have any relevant financial relationships to be discussed, directly or indirectly, referred to or illustrated with or without recognition within the presentation.
I have no financial relationships beyond my employment at Intermountain Healthcare.
Quality, Utilization, & Efficiency (QUE) Six clinical areas studied over 2 years:- transurethral prostatectomy (TURP)- open cholecystectomy- total hip arthroplasty- coronary artery bypass graft surgery (CABG)- permanent pacemaker implantation- community-acquired pneumoniapulled all patients treated over a defined time period
across all Intermountain inpatient facilities - typically 1 yearidentified and staged (relative to changes in expected utilization)- severity of presenting primary condition- all comorbidities on admission- every complication- measures of long term outcomescompared physicians with meaningful # of cases
(low volume physicians included in parallel analysis, as a group)
Intermountain TURP QUE StudyMedian Surgery Minutes vs Median Grams Tissue
M L K J P B C O N A I D H E G F0
20
40
60
80
100
0
20
40
60
80
100
Attending Physician
Median surgical time Median grams tissue removed
Gra
ms
tissu
e / S
urge
ry m
inut
es
Intermountain TURP QUE Study
1500 1549 1568 16181543
1697
1913
22332140 2156
1598
12691164
1552 15561662
A B C D E F G H I J K L M N O PAttending Physician
0
500
1000
1500
2000
2500
Dol
lars
0
500
1000
1500
2000
2500
Average Hospital Cost
Total Hip Arthroplasty - LOS
1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3
0
2
4
6
8
10
12
14
16
Leng
th o
f Sta
y (d
ays)
1988 1989 1990Month/Year
0
2
4
6
8
10
12
14
161 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3
1988 1989 1990
Total Hip Arthroplasty - Cost
0
2
4
6
8
10
12
14
1988 1989 1990
0
2
4
6
8
10
12
141 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3
1988 1989 1990
Month/Year
Ave
rage
cos
t per
cas
e ($
1,00
0s)
1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3
Deming: Quality controls cost
Quality Cost Forum
internal
internal
Cost-benefit society
-
Waste:Savings Potential
25-40%
> 50%
(none)
Inefficiency waste
Quality waste
Deep post-op wound infections
% prophylaxis givenat optimal time
1985 1986 1991
40 58 96
% Infections 1.8 0.9 0.4
Est. decrease in infectionsrelative to 1985 rate -- 33 51
Est. savings at $14,000per case (in thousands) -- 462 714
National standard: 2 - 4% deep post-op wound infection rate
LDSH Dept of Clinical Epidemiology
Deep post-op wound infections 1985 1994
38.0 37.1
40.0 99.1
19.0 5.3
43.0 14.3
% elective surgeriesreceiving prophylaxis
% receiving first dose0-2 hrs before incision
% continuing prophylaxis24 hrs after surgery
Mean number ofdoses per case
LDSH Dept of Clinical Epidemiology
NIH-funded randomized controlled trialassessing an "artifical lung" vs. standard ventilator managementfor acute respiratory distress syndrome (ARDS)
discovered large variations in ventilator settings across and within expert pulmonologists
created a protocol for ventilator settings in the control arm of the trial
Dr. Alan Morris, LDS Hospital, 1991:We generalized the method
Problems with "best care" protocols
Lack of evidence for best practice- Level 1, 2, or 3 evidence available only about 15-25% of the time
Expert consensus is unreliable- experts can't accurately estimate rates using subjective recall
(produce guesses that range from 0 to 100%, with no discernable pattern of response)- what you get depends on whom you invite (specialty level, individual level)
Guidelines don't guide practice- systems that rely on human memory execute correctly ~50% of the time (McGlynn: 55% for adults, 46% for children)
No two patients are the same; therefore, no guideline perfectly fits any patient (with very rare exception)
NIH-funded randomized controlled trialassessing an "artifical lung" vs. standard ventilator managementfor acute respiratory distress syndrome (ARDS)
discovered large variations in ventilator settings across and within expert pulmonologists
created a protocol for ventilator settings in the control arm of the trial
Implemented the protocol using Lean principles (Womack et al., 1990 - The Machine That Changed the World)- built into clinical workflows - automatic unless modified- clinicians encouraged to vary based on patient need- variances and patient outcomes fed back in a Lean Learning Loop
Dr. Alan Morris, LDS Hospital, 1991:We generalized the method
1. Identify a high-priority clinical process (key process analysis)
2. Build an evidence-based best practice protocol(always imperfect: poor evidence, unreliable consensus)
3. Blend it into clinical workflow (= clinical decision support; don't rely on human memory; make "best care" the lowest energy state, default choice that happens automatically unless someone must modify)
4. Embed data systems to track (1) protocol variations and (2) short and long term patient results (intermediate and final clinical, cost, and satisfaction outcomes)
5. Demand that clinicians vary based on patient need6. Feed those data back (variations, outcomes) in a Lean
Learning Loop- constantly update and improve the protocol- provide true transparency to front-line clinicians- generate formal knowledge (peer-reviewed publications)
Shared Baseline "Lean" protocols (bundles)
ARDS Protocol Compliance
29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 590
10
20
30
40
50
60
70
80
90
100
ARDS Patient Number
% P
roto
col I
nstr
uctio
ns F
ollo
wed
0
10
20
30
40
50
60
70
80
90
10029 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59
Results:survival (for ECMO entry criteria patients) improved from 9.5% to 44%costs fell by ~25% (from $160k to $120k)physician time fell by ~50%
Dr. Alan Morris, LDS Hospital, 1991
Key take-aways
1. No protocol perfectly fits any patient- solution: Shared Baseline "bundles"
(mass customization = "patient centered care")
2. Serious limitations to protocol development- solution: a Learning System (embedded variance and outcomes
tracking; continuous protocol review and tested improvement)
3. Reliance on human memory (craft of medicine) produces "55% execution"- solution: tools to embed protocols in workflows
4. Only two differences from traditional practice: It requires (1) coordinated teams with (2) reliable data systems
there is nothing new here ...
It should have started in medicine ...
except the idea that
"it takes a team"(and true transparency = embedded data systems)
07 Ja
nMar May Ju
lSep Nov
08 Ja
nMar May Ju
lSep Nov
09 Ja
nMar May Ju
lSep Nov
10 Ja
nMar
Month
0
20
40
60
80
100
% c
ompl
ianc
e
0
20
40
60
80
100
ER bundle ICU bundle All components
Sepsis bundle compliance
04 Ja
nMay Sep
05 Ja
nMay Sep
06 Ja
nMay Sep
07 Ja
nMay Sep
08 Ja
nMay Sep
09 Ja
nMay Sep
10 Ja
n
Month
0
0.1
0.2
0.3
0.4
0.5
Mor
talit
y ra
te
0
0.1
0.2
0.3
0.4
0.5
Sepsis mortality - ER-ICU transfers
20.2%
8.0%
~116 fewer inpatient deaths per year
2832
4437
4542
4223
3429
4133
4553
3850
4739
3130
3424
4041
3528
2722
2827
2432
4436
3952
5170
6560
4757
5250
6151
4377
7377
6571
6948
5259
4663
6868
6370
9490
7581
6979
8178
8270
7484
91n=
We count our successes in lives ...
Lesson 1
6.66
3.36
2.47 2.65
3.44
4.26
37 38 39 40 41 42
Weeks gestation
0
2
4
6
8
10
Perc
ent N
ICU
adm
issi
ons
0
2
4
6
8
10
Deliveries w/o Complications, 2002 - 2003
8,001 18,988 33,185 19,601 4,505 258n =
NICU admits by weeks gestation
Elective inductions < 39 weeks
5.55.1
6.66.3 65.3
8.2
5.45.76.66.6
7.9
6.4
7.67.6
4.63.5
4.54.3
6.5
3.22.62.3
4.2
2.13.23.4
2.4
5
33.5
26.726.9
2929.2
25.3
27.6
20.4
19.1
16.5
15.2
8.4
10.7
8.1
6.85.96.1 6
5.1
6.3
Jan 01 Mar May Ju
lSep Nov
Jan 02 Mar May Ju
l
Jan 03 Mar May Ju
lSep Nov
Jan 04 Mar May Ju
lSep Nov
Jan 05 Mar May Ju
l
0
5
10
15
20
25
30
% e
lect
ive
indu
ctio
ns <
39
wee
ks
0
5
10
15
20
25
30
382372
490415
430435
422455
430382
356337
372366
455n = 423453
473476 512
475602
557667
564637
578541
573533
505501
474536
562545
535493
520494
430440
500421
474562
549555
528491
3331.4
36.1
28.3
17.7
15.1
17.6
14.4 14.3
5.84.5
2.1
0
20
8.2 8.5
3.6 3.4 3.9 3.22.4
1.1 0.9 10 0
1 2 3 4 5 6 7 8 9 10 11 12 13
Bishop score
0
5
10
15
20
25
30
35
40
Perc
ent c
-sec
tions
0
5
10
15
20
25
30
35
40
Unplanned c-section ratesElectively induced patients by Bishop score, Jan 2002 - Aug 2003
10 49 130 274 567 856 1114 1266 1062 737 415 86 1918 35 61 99 164 278 375 487 453 346 179 47 7
MultipsPrimips
n
22.1
20.7
17.4
15.715
13.8
12.611.6
10.4
9 9
7.58.2
12.4 12
10.810.1
9.28.1
7.67.1
6.45.9 5.5 5.1
4.1
1 2 3 4 5 6 7 8 9 10 11 12 13
Bishop score
0
5
10
15
20
25
Hou
rs
0
5
10
15
20
25
Average hours in labor & deliveryElectively induced patients by Bishop score, Jan 2002 - Aug 2003
10 49 130 274 567 856 1114 1266 1062 737 415 86 1918 35 61 99 164 278 375 487 453 346 179 47 7
MultipsPrimips
n
15.314
15.314.514.7
11.612.8
11.812.612.8
15.1
12.19.9
8.86.8 6.5 6 6.1
7.66.5 6.6
5.2 4.9
8.4
4.3 4.3 4.56.1 5.4
4.4 3.9
53 53
63
5357
45
5652
41
52
62
4649
35
21 2126 28
3428
2218 20
35
1518
1518
2521 20
110
87
119
109
124
91
107
94100
105
118
8781
67
57 57
4652
6055
49
3733
67
30 3036
4845
3734
Jan 20
03 Feb Mar AprMay Ju
n Jul
AugSep OctNovDec
Jan 20
04 Feb Mar AprMay Ju
n Jul
AugSep OctNovDec
Jan 20
05 Feb Mar AprMay Ju
n Jul
0
20
40
60
80
100
120
140
Num
ber o
f pat
ient
s
0
10
20
30
40
50
% o
f all
prim
ipar
ous
deliv
erie
s
Primiparous elective inductions
Bishop's score < 10Bishop's score < 8Goal: Reduce "inappropriate" nullip inductions by 50%
Elective induction: length of labor
Jan 20
01 Mar May Jul
Sep NovJa
n 2002 Mar May Ju
lSep Nov
Jan 20
03 Mar May Jul
Sep NovJa
n 2004 Mar May Ju
lSep Nov
Jan 20
05 Mar May Jul
Sep Nov
0
2
4
6
8
10
Ave
rage
hou
rs fr
om a
dmis
sion
to d
eliv
ery
0
2
4
6
8
10
8.5
7.97.5
7.16.9
(note: includes all elective inductions)
Overall c-section rate
96 97 98 99
2000 01 02 03 04 05 06
0%
10%
20%
30%
40%
Perc
ent c
-sec
tions
ove
rall
0%
10%
20%
30%
40%
National Intermountain
2001 2002 2003 20040
2,000,000
4,000,000
6,000,000
8,000,000
10,000,000
Cos
t str
uctu
re im
prov
emen
t ($)
0
1,000,000
2,000,000
3,000,000
4,000,000
5,000,000
6,000,000
7,000,000
8,000,000
9,000,000
10,000,000
Cum
ulat
ive
annu
al to
tal (
$)
Combined maternal and neonatal variable costDeliveries without complications resulting in normal newborns
Actual - expected cost, based on year-end 2000 with PPI inflation
Quality-based cost improvement
Very often,
better care is cheaper care ...
Lesson 2
50+% of all resource expenditures in hospitals is
quality-associated waste:recovering from preventable foul-upsbuilding unusable productsproviding unnecessary treatmentssimple inefficiency
Andersen, C. 1991James BC et al., 2006
No good deed goes unpunishedNeonates > 33 weeks gestational age
who develop respiratory distress syndromeTreat at birth hospital with nasal CPAP (prevents
alveolar collapse), oxygen, +/- surfactantTransport to NICU declines from 78% to 18%.Financial impact (NOI; ~110 patients per year; raw $):
Birth hospitalTransport (staff only)
Tertiary (NICU) hospitalDelivery system total
Integrated health planMedicaid
Other commerical payersPayer total
Before 84,24422,199
958,4671,064,910
900,599652,103
429,1011,981,803
After 553,479
- 27,222 209,829
736,086
512,120373,735
223,2151,109,070
Net 469,235
- 49,421 -748,638
-328,824
388,479278,368
205,886872,733
Current U.S. payment mechanisms
Actively incent overutilization: do more, get paid more - even when there is no health benefit
I am paid to harm my patients (paid more for complications)
Actively disincents innovation that reduces costs through better quality (a key success factor for the rest of the U.S. economy)
Very strong, deep, wide evidence showing exactly this effect throughout U.S. healthcare
1. ACOs, AMHs, bundled payment, shared savings, pay for value: sophisticated forms of capitation- provider at (financial) risk ... but with far better data systems for
(1) quality measurement and (2) risk adjustment
2. Represent "managed care at the bedside"- ask clinical teams at the bedside to manage the care, not distant
and disengaged insurance companies
3. More than 80% of cost saving opportunities live on the clinical side; 70+% of clinical improvement activities reduce costs by freeing up care delivery capacity (technically, "fixed cost leverage").
Capitation makes a comeback
A fundamental shift in focusThe past:
1. "Top-line" revenue enhancement- Systems designed around documentation to support FFS payment,
clinical decision support as a secondary "bolt-on"
2. Quality defined as regulatory compliance - e.g.- CMS Core Measures- Pay for Value- Meaningful Use
The future:1. Quality becomes the core business
- Demonstrated performance for key clinical processes- Systems designed around clinical decision support (process
management), producing documentation as an integrated by-product
2. "Bottom-line" cost control and waste eliminationin a "provider at risk" financial environment
1. Quality improvement is the science of process management
2. A focus on process management forces patient-centered care - care built along the full
continuum of care; not buildings, technologies, or physicians
3. Combining patient- centered care with various levels of provider-at-financial-risk forces population-level care and the triple aim
- collaborating with other community organizations (churches, schools, local governments, etc.) to promote best health and high function
Getting to the Triple Aim
Better has no limit ...
an old Yiddish proverb