An RCT of a clinician communication training intervention to improve - - PowerPoint PPT Presentation

An RCT of a clinician communication training intervention to improve pain management in primary care

Stephen Henry, MD MSc CHPR seminar February 5, 2020

Disclosures

None

Objectives

• 1. Review “best practices” for communicating about

pain and opioids in primary care

• 2. Describe benefits of using standardized patient

instructors to change clinician behavior

• 3. Identify impacts of communication training on

communication during visits and pain-related care

• utcomes

Background

• Discussions about chronic pain are
• ften frustrating for both patients

and clinicians

• Clinicians often struggle to

manage patients taking opioids for chronic pain

• Communication skills training may

result in more satisfied patients, less frustrated physicians, and better pain management

Project overview

Interviews with patients, residents, clinical experts Develop communication framework Design intervention to teach framework Conduct pilot RCT to test framework

Project overview

Interviews with patients, residents, clinical experts Develop communication framework Design intervention to teach framework Conduct pilot RCT to test framework

Qualitative data collection : 1-on-1 interviews: 15 patients with chronic pain 10 UCD primary care residents Conversations with clinical experts Focus group data from tapering project Video-recorded primary care visits involving patients with pain

Project overview

Interviews with patients, residents, clinical experts Develop communication framework Design intervention to teach framework Conduct pilot RCT to test framework

Step 1: Mentally prepare for the visit Step 2: Show that you take the patient’s pain seriously Step 3: Assess the patients’ risk for opioid- related harm Step 4: Set pain treatment goals Step 5: Develop a goal-directed treatment plan

Project overview

Interviews with patients, residents, clinical experts Develop communication framework Design intervention to teach framework Conduct pilot RCT to test framework

Used “Standardized Patent Instructor” approach to delivering this behavioral intervention. JGIM 2009; 24:606-13 (SEE-IT residents) JGIM 2016;31:715-722 (SEE-IT PCPs) JAMA IM 2016;176:191-7 (PCORI - residents) JAMA Oncology 2017;3:92-100 (VOICE-

• ncologists)

Intervention components

SP visit #1

~ 9 minute instructional video Pamphlet with detailed examples Pocket card

SP#1 role play and feedback SP#2 role play and feedback

Project overview

Interviews with patients, residents, clinical experts Develop communication framework Design intervention to teach framework Conduct pilot RCT to test framework

RCT overview

Recruit physicians Clinician intervention

R

Clinician control Recruit patients (1-2 per MD) Audio record clinic visits Post-visit surveys 2-month follow up phone call EHR data Code visit recordings

Recruitment - physicians

• IM or FCM residents at UC Davis
• PGY2 or greater
• Have continuity clinic in ACC

Recruitment - patients

INCLUSION

• Established adult clinic patients
• On long-term opioids (≥1 dose per day for >90 days)
• Chronic MSK pain with severity ≥4 out of 10
• Scheduled to see an enrolled physician
• Report either dissatisfaction with current pain meds
• r desire to discuss change in pain meds at visit

Recruitment - patients

EXCLUSION

• Doesn’t speak English during visits
• Active cancer treatment
• Palliative care / hospice
• Opioids not managed by primary care clinic

Physician control

CDC handout on opioid prescribing and pain management

Outcomes

ClinicalTrials.gov ID: NCT03629197

Primary

• # of targeted communication skills coded from

recorded patient visits

• Pain-related interference 2 months after study visit

Secondary

• Physician self-efficacy for communication and pain

management

• Physician reported visit difficulty
• Patient experience
• Pain intensity 2 months after study visit

Visit coding

• Identified observable behaviors for each step in the 5

step framework

Visit coding

Framework step Step 1: Mentally prepare for the visit Step 2: Show that you take the patient’s pain seriously Step 3: Assess the patients’ risk for

• pioid-related harm

Step 4: Set pain treatment goals Step 5: Develop a goal-directed treatment plan

Visit coding

Framework step Observable behaviors (26 total) Step 1: Mentally prepare for the visit None (0) Step 2: Show that you take the patient’s pain seriously Open-ended questions (5) Supportive / empathetic statements (3) Elicit patient opinion (1) Step 3: Assess the patients’ risk for

• pioid-related harm

Ask about opioid side effects (2) Step 4: Set pain treatment goals Setting function-based goals (4) Step 5: Develop a goal-directed treatment plan Patient input into treatment (3) Responding to tapering resistance (2) Flexibility and follow up (4)

Visit coding

• 3 coders trained to code from transcripts and audio
• 3 coders independently coded each transcript
• Disagreements resolved by discussion

Physician participants

CDC guidelines (n=21) Intervention (n=24) Total (n=45) Age, mean (SD) 29.8 (2.7) 29.6 (2.5) 29.7 (2.5) Sex, n (%) Female 8 (38%) 21 (88%) 29 (64%) Male 13 (62%) 3 (13%) 16 (36%) Race/ethnicity, n (%) Black 1 (5%) 0 (0%) 1 (2%) Asian / Pac Island. 12 (57%) 9 (38%) 21 (47%) Non-Hispanic White 5 (24%) 7 (29%) 12 (27%) Hispanic White 2 (10%) 4 (17%) 6 (13%) Multi / other 1 (5%) 4 (17%) 5 (11%) Clinic*, n (%) Internal Medicine 14 (67%) 16 (67%) 30 (67%) Family Medicine 7 (33%) 8 (33%) 15 (33%) Self-efficacy for pain- related communication†, mean (SD) 3.05 (0.57) 2.83 (0.46) 2.94 (0.52)

*Randomization was stratified by clinic † Mean of 8 different items; range 1-5 (higher = greater self-efficacy).

Patient participants

Saw control MD (n=21) Saw intervention MD (n=25) Total (n=46) Age, mean (SD) 59.9 (8.6) 61.2 (10.5) 60.6 (9.6) Female sex, n (%) 12 (57%) 20 (80%) 32 (70%) Race/ethnicity, n (%) Black 7 (33%) 9 (36%) 16 (35%) Non-Hispanic White 8 (38%) 9 (36%) 17 (37%) Native American 1 (5%) 0(0%) 1 (2%) Hispanic White 2 (10%) 4 (16%) 6 (13%) Multi / other 3 (14%) 3 (12%) 6 (13%) Household income, n (%) <$10,000 4 (19%) 8 (32%) 12 (26%) $10,001 – $20,000 8 (38%) 5 (20%) 13 (28%) $20,001 – $40,000 5 (24%) 6 (24%) 11 (24%) $40,001 – $80,000 3 (14%) 3 (12%) 6 (13%) >$80,000 1 (5%) 3 (12%) 4 (9%) Employment status*, n (%) Working full time 2 (10%) 2 (8%) 4 (9%) Disabled / unable to work 10 (48%) 15 (60%) 25 (54%) Retired 5 (24%) 6 (24%) 11 (24%) *2 small categories not shown

Patient participants

Saw control MD (n=21) Saw intervention MD (n=25) Total (n=46) Clinic, n (%) Internal Medicine 11 (52%) 15 (60%) 26 (57%) Family Medicine 10 (48%) 10 (40%) 20 (43%) Saw usual MD, n (%) 14 (67%) 21 (84%) 35 (76%) Brief Pain Index Pain-related interference (0-10) 6.2 (2.5) 7.4 (1.9) 6.9 (2.2) Pain-related intensity (0-10) 6.4 (1.8) 7.3 (1.8) 6.9 (1.8) PHQ-8 (0-24) 10.6 (6.1) 10.0 (6.8) 10.3 (6.4) GAD-7 (0-21) 8.0 (5.0) 9.6 (6.4) 8.9 (5.8) Opioid dose, median MME/day, (IQR) 22 (15, 40) 30 (15, 45) 30 (15, 45)

• No. of pain locations, median (IQR)

2 (2,3) 2 (1-2) 2 (1, 3)

Results

Outcome Saw control MD (n=21) Saw intervention MD (n=25) Physician communication Total communication behaviors coded per visit, mean (SD) Post-visit measures Patient experience* MD-reported visit difficulty (10-60) MD self-efficacy for pain-related communication (1-5)** 2-month follow up measures Pain interference (0-10) Pain intensity (0-10) *standardized variable (mean 0, SD 1); higher values = more positive **mean of 8 items (each scored 1-5)

Results

Outcome Saw control MD (n=21) Saw intervention MD (n=25) Physician communication Total communication behaviors coded per visit, mean (SD) 2.8 (2.4) 3.5 (2.8) Post-visit measures Patient experience* MD-reported visit difficulty (10-60) MD self-efficacy for pain-related communication (1-5)** 2-month follow up measures Pain interference (0-10) Pain intensity (0-10) *standardized variable (mean 0, SD 1); higher values = more positive **mean of 8 items (each scored 1-5)

Results

Outcome Saw control MD (n=21) Saw intervention MD (n=25) Physician communication Total communication behaviors coded per visit, mean (SD) 2.8 (2.4) 3.5 (2.8) Post-visit measures Patient experience* 0.13 (0.6)

• 0.12 (1.1)

MD-reported visit difficulty (10-60) 22.8 (7.0) 25.8 (12.3) MD self-efficacy for pain-related communication (1-5)** 3.4 (0.6) 3.8 (0.6) 2-month follow up measures Pain interference (0-10) Pain intensity (0-10) *standardized variable (mean 0, SD 1); higher values = more positive **mean of 8 items (each scored 1-5)

Results

Outcome Saw control MD (n=21) Saw intervention MD (n=25) Physician communication Total communication behaviors coded per visit, mean (SD) 2.8 (2.4) 3.5 (2.8) Post-visit measures Patient experience* 0.13 (0.6)

• 0.12 (1.1)

MD-reported visit difficulty (10-60) 22.8 (7.0) 25.8 (12.3) MD self-efficacy for pain-related communication (1-5)** 3.4 (0.6) 3.8 (0.6) 2-month follow up measures Pain interference (0-10) 5.8 (2.9) 6.3 (2.8) Pain intensity (0-10) 6.0 (1.8) 6.5 (2.3) *standardized variable (mean 0, SD 1); higher values = more positive **mean of 8 items (each scored 1-5)

Outcome Effect for Intervention Physician communication, (IRR, 95%CI)* Total communication behaviors coded per visit Post-visit measures, (coef.,95%CI)** Patient experience MD-reported visit difficulty (10-60) MD self-efficacy for pain-related communication† (1-5) 2-month follow up measures** Pain interference† (0-10) Pain intensity† (0-10) *Poisson regression (GLM) using GEE to account for clustering **Linear regression (GLM) using GEE to account for clustering †Models control for baseline values

Results

Outcome Effect for Intervention Physician communication, (IRR, 95%CI)* Total communication behaviors coded per visit 1.28 (0.76, 2.13) Post-visit measures, (coef.,95%CI)** Patient experience MD-reported visit difficulty (10-60) MD self-efficacy for pain-related communication† (1-5) 2-month follow up measures** Pain interference† (0-10) Pain intensity† (0-10) *Poisson regression (GLM) using GEE to account for clustering **Linear regression (GLM) using GEE to account for clustering †Models control for baseline values

Results

Outcome Effect for Intervention Physician communication, (IRR, 95%CI)* Total communication behaviors coded per visit 1.28 (0.76, 2.13) Post-visit measures, (coef.,95%CI)** Patient experience

• 0.25 (-0.76, 0.27)

MD-reported visit difficulty (10-60) 3.04 (-2.23, 8.32) MD self-efficacy for pain-related communication† (1-5) 0.42 (0.12, 0.72) 2-month follow up measures** Pain interference† (0-10) Pain intensity† (0-10) *Poisson regression (GLM) using GEE to account for clustering **Linear regression (GLM) using GEE to account for clustering †Models control for baseline values

Results

Outcome Effect for Intervention Physician communication, (IRR, 95%CI)* Total communication behaviors coded per visit 1.28 (0.76, 2.13) Post-visit measures, (coef.,95%CI)** Patient experience

• 0.25 (-0.76, 0.27)

MD-reported visit difficulty (10-60) 3.04 (-2.23, 8.32) MD self-efficacy for pain-related communication† (1-5) 0.42 (0.12, 0.72) 2-month follow up measures** Pain interference† (0-10)

• 0.30 (-1.7, 1.17)

Pain intensity† (0-10)

• 0.27 (-1.15, 0.62)

*Poisson regression (GLM) using GEE to account for clustering **Linear regression (GLM) using GEE to account for clustering †Models control for baseline values

Results

Discussion

• A physician communication training intervention did

not result in significantly higher rates of targeted pain- related communication behaviors in subsequent visits

• The training significantly improved physician self-

efficacy for pain-related communication

• No differences in patient experience, or physician

perception of visit difficulty

• No difference in patient pain 2 months after visit

Exploratory analysis

Framework step Observable behaviors (26 total) Step 1: Mentally prepare for the visit None (0) Step 2: Show that you take the patient’s pain seriously Open-ended questions (5) Supportive / empathetic statements (3) Elicit patient opinion (1) Step 3: Assess the patients’ risk for

• pioid-related harm

Ask about opioid side effects (2) Step 4: Set pain treatment goals Setting function-based goals (4) Step 5: Develop a goal-directed treatment plan Patient input into treatment (3) Responding to tapering resistance (2) Flexibility and follow up (4)

Exploratory analysis

Outcome Saw control MD (n=21) Saw intervention MD (n=25) Physician communication Total communication behaviors coded per visit, mean (SD) 2.8 (2.4) 3.5 (2.8) Total step 2 behaviors 1.1 (1.0) 2.4 (1.5)* Total step 3 behaviors 0.22 (0.55) 0.38 (0.74) Total step 4 behaviors 0.1 (0.4) 0.3 (0.6) Total step 5 behaviors 1.2 (1.2) 0.7 (1.1) * IRR 2.2 (95%CI 1.34, 3.61) P = 0.003

Potential next steps

• Better measures of intervention effect?
• Additional / modified outcomes
• Modify intervention

– Add components

• Appropriate population (PCPs vs residents)

Results

Disagree/ Str. Disagree Neutral Agree/ Str. Agree The training I received will be helpful in my practice 24 (100%) The training was easy to understand 24 (100%) The actors’ portrayals of patients with chronic pain were realistic 1 (4%) 3 (13%) 20 (83%) The training was relevant to my daily clinical practice 24 (100%) I will use the knowledge and/or skills I gained form the training during visits 24 (100%) This project will lead to improved care for my patients 24 (100%) The study disrupted patient flow 15 (63%) 7 (29%) 2 (8%) Participating in the study was a hassle 18 (75%) 5(21%) 1 (4%) If asked, I would agree to participate in a similar study in the future 2 (8%) 22 (92%) 100% of intervention residents rated training quality good or excellent

Acknowledgements

Funding NIH grant K23DA043052, UCD Dept of Internal Medicine Project manager & SP trainer Gary Weinberg Faculty mentors Rich Kravitz, Josh Fenton, Cynthia Campbell (KP), Mark Sullivan (UW) RAs Hiba Naz, Sherry Hao Coders Gary Weinberg, Hiba Naz, Wyatt Graham CHPR staff, SPs, residents