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