The Chest Pain Choice Decision Aid: a Randomized Trial ISDM Conference Maastricht, June 2011.

The Chest Pain Choice Decision Aid: a Randomized

Trial

ISDM Conference

Maastricht, June 2011

E Hess, M. Knoedler, N. Shah, J Kline, M Breslin, M Branda, L Pencille, B Asplin, D Nestler, A Sadosty, H. Ting, M. Montori

Knowledge and Evaluation Research Unit

Mayo Clinic College of Medicine

MNFoundation for Informed Medical

Decision Making

AHA Fellow-to-Faculty Transition Award

Background• Chest pain 2nd most common

complaint in U.S. Emergency Departments

> 6 million patients annually

• 4% of ACS inappropriately discharged from ED

• Large #’s of low risk patients admitted for prolonged observation and cardiac stress testing

• False positive test results, unnecessary procedures, cost

Pope, NEJM, 2000

Background

• Kline and colleagues developed a quantitative pretest probability calculator

• Prospectively validated QPTP calculator in 3 Academic EDs

• Demonstrated efficacy of QPTP calculator in RCT

4

Kline JA, BMC Med Informed Decision Making, 2005

Kline JA, Annals of Emergency Medicine, 2009

Mitchell AM, Kline JA, Annals of Emergency Medicine, 2006

Background

Decision Aids: knowledge (by 15 of 100, 95% confidence interval 12-19%)% patients with realistic

perceptions of the chances of benefits and harms by 60% (40-90%) uncertainty related to feeling uninformed (by 8 of 100 (5-12)% passive patients in decision making by 30% (10-50%)% remaining undecided after counseling by 57% (30-70%)

O’Connor, Cochrane Database of Systematic Reviews, 2009

Hypothesis

Facilitating a patient-centered discussion regarding the short-term risk for ACS in otherwise low-risk chest pain patients will:

patient knowledge

patient engagement

Safely resource use

Objectives

(1) To design a DA for use in patients at low risk for ACS

(2) To test the DA in a randomized trial

Methods

Decision Aid Design

• Incorporate QPTP output in a literacy-sensitive DA, describe rationale of evaluation, list management options in value-neutral fashion

• Iteratively test DA in patient encounters

• Refine DA based on input from patients, clinicians, and investigative team thematic saturation

Breslin, Mullan, Montori Patient Educ Counseling 2008

Methods: Clinical Trial

• Design: single-center; allocation concealed by password-protected, web-based randomization

• Setting: Academic ED in Rochester, MN with 73,000 annual patient visits; 10-bed observation unit

• Eligibility:–Included: Adults with chest pain considered for EDOU

admission–Excluded: +troponin, known CAD, cocaine use within

72 hrs, unable to provide informed consent or use decision aid

Outcome measures

• Decision quality–Patient knowledge**–Degree of patient participation (OPTION scale)–Decisional conflict (DCS)–Trust in physician (TPS)

• Quantitative–Safety endpoint: 30-day MACE*–Resource use

• Rate of cardiac stress testing in EDOU• 30-day rate of stress testing

Statistical analysis

• Power: 200 patients–90% power to detect > 25% ↑ in mean knowledge–95% power to detect a 20% ↓ in proportion of

patients who underwent stress testing in EDOU

• Hypothesis testing: chi-square, Fisher’s exact, t-test or Wilcoxon rank-sum as appropriate

• Intention-to-treat principle followed

Results

Baseline Characteristics

Variable Intervention

(n=101)

Control

(n=103)

P-value

Mean age 54.5 54.9 0.81Female 59% 61% 0.97HTN 45% 28% 0.01Hyperlipidemia 45% 39% 0.46Family history of premature CAD

14% 12% 0.61

Mean PTP of ACS 3.2% 3.3% 0.81

Knowledge and Participation

Variable Intervention

(n=101)

Control

(n=103)

Mean diff (95% CI)/ p-value

6 knowledge questions

3.6 3.0 0.67

(0.34, 1.0)OPTION score 51.4 32 < 0.0001

Decisional Conflict* and Physician Trust

Variable Intervention

(n=101)

Control

(n=104)

Mean diff (95% CI)

Decisional conflict (DCS)

22.3 43.3 -13.6

(-19.1, -8.1)Trust in physician (TPS)

83.4 79.3% 4.1

(-1.4, 9.6)

*Conflict related to feeling uninformed

Acceptability to Patients

Variable Intervention

(n=101)

Control

(n=104)

P-value

Amount of information

(just right)

93% 80% 0.0051

Clarity of information (extremely clear)

62% 37% <0.0001

Helpfulness (extremely helpful)

53% 34% <0.0001

Would recommend to others

75% 45% <0.0001

Provider experience

Variable Intervention

(n=101)

Control

(n=104)

P-value

Strongly recommend way information was shared

59% 20% <0.0001

Want to present other diagnostic information in same way

64% 28% <0.0001

Safety

Variable Intervention

(n=101)

Control

(n=104)

P-value

Revascularization 3% 2% 0.68MI 1% 0% 0.49Death 0 0 NAMACE within 30 days of discharge

0 0 NA

Resource use

Variable Intervention

(n=101)

Control

(n=104)

P-value

Stress test in EDOU

58% 77% <0.0001

Stress test performed within 30 days

75% 91% 0.02

Follow-up as outpatient

39% 9% <0.0001

Limitations

• Single center

• Insufficient power to demonstrate safety

Conclusions

Summary of impact of DA

Variable Direction of differencePatient knowledge ↑Patient participation

↑

Decisional conflict ↓Physician Trust ↔Acceptability ↑Safety ↔Resource use ↓

Lessons learned

• Integration in process of care challenging

• Care process redesign required??

• Feasibility of definitively demonstrating patient safety?

• Use of DA in emergency department requires reliable access to outpatient follow-up

Future Directions

• Identification of factors that promote or inhibit uptake of SDM in acute setting

• Prospective multicenter randomized trial