The Wisdom of Crowds: Can Online Clinician Engagement Impact Health Technology Assessment?

Select to access the PowerPoint® presentation (645 KB).

Slide 1

The Wisdom of Crowds: Can Online Clinician Engagement Impact Health Technology Assessment?

Carl N. Kraus, MD
Vice President, Medical Affairs
Medscape
September 10, 2012

2012 AHRQ Annual Conference

Disclaimer: Any views or opinions presented here are solely those of myself and do not necessarily represent those of Medscape. The examples included in this presentation are intended for discussion purposes only. I have no financial relationship with anti-obesity manufacturers.

Slide 2

Outline

• Crowdsourcing and online clinician content consumption.
• Stakeholder disparity in health technologies.
• Case Study: Obesity.

Slide 3

Background

• 2004: James Surowiecki—thesis is that independently deciding individuals (groups) can make better decisions and predictions than individuals/experts.
• There are differences between "wise" crowds and "irrational" crowds:
  • Diversity of opinion.
  • Independence of opinion.
  • Decentralization.
  • Ability to aggregate opinion.

Image: The covers of two books, The Wisdom of Crowds, by James Surowiecki, and Crowd Sourcing, by Jeff Howe, are shown.

Slide 4

Medscape's Role in Content Consumption

• 2.6 million U.S. physician visits per month.
• 33 distinct specialty sites.
• 539,000 12-month active U.S. physicians.
• 312,000 U.S. Physicians using Medscape via Mobile.

Image: A blue figure in a white coat, representing a physician, is shown. Three arrows encompassed by a red circle, point away from this figure. The first arrow passes the following text, "Oath of care: Improve patient care" and a small photograph of a doctor speaking with a patient, then points to the physician figure with another blue figure with its leg, arm, and head in bandages. The second arrow passes the following text, "REMS: Access to therapies" and the Food and Drug Administration (FDA) logo, then points to a small photograph of two blue and red capsules. The third arrow passes the following text, "MoL, MoC: Ability to practice" and a small photograph of a building captioned "Duke Primary Care", then points to a blue figure standing before an easel with a dollar sign on it.

Slide 5

Medscape Environment as a Platform for "Crowdsourcing?"

• Diversity of opinion:
  • Multiple specialties.
  • Multiple professions.
  • Multiple geographic locales.
• Independence of opinion:
  • Response of one participant does not impact that of another.
• Decentralization:
  • Experience of each individual influences responses locally.
• Ability to aggregate opinion:
  • Data capture is electronically structured and can be subsequently analyzed.

Conclusion: leveraging the opinion of clinicians "may" provide insight into the future value of health technologies.

Slide 6

Primary Stakeholders in Health Technology Adoption

Image: The figure shows a blue arrow pointing across the slide from left to right; along its length is the following text, accompanied by blue dots: Ex vivo (1 blue dot), -omics (2 blue dots), FIH (3 blue dots), PP vs. ITT (4 blue dots), Superiority vs. NI, BR Comm. (6 blue dots).

Below the arrow is the following text in three boxes:

• Technology developers:
  • Biotech.
  • Pharma.
  • Device manufacturers.
  • Application developers.
• Technology assessors:
  • CDER.
  • CBER.
  • CDRH.
  • CMS.
  • AHRQ/NICE.
  • Insurers.
  • Venture Capital.
  • CROs (risk sharing).
• Technology users:
  • Physicians.
  • Patients.
  • Caregivers.
  • Nurse practitioners.
  • Physician assistants.
  • Pharmacists.

Green arrows point down from "Technology developers" and "Technology assessors" to a box across the bottom of the slide, containing the following text: "What 'developer/assessor' variables might be of impact to the user community?" A red arrow points up from this text box to "Technology users."

Slide 7

Why Opinions Vary

21CFR314.126 (new drug regulations):

• FDA's role is to determine whether "an investigation is adequate and well-controlled... the primary basis for determining whether there is 'substantial evidence' to support the claims of effectiveness for new drugs."

Credible accessible data for analysis:

• Pharm/tox.
• Clin/pharm.
• Efficacy.
• Safety.

Modified Oath of Geneva:

• "The health and life of my patient will be my first consideration; I will maintain by all means in my power, the honor and the noble traditions of the medical profession."

Credible accessible data for analysis:

• Curbside information.
• Product label.
• Peer reviewed clinical studies.
• CME.
• Promotional education.

Slide 8

Why Clinicians Face Learning Curve Constraints on New Technologies

1. Follow-up from last visit (lab results, films, consults).
2. Interval changes.
3. Stratify diagnostic/therapeutic interventions based on CC/PMH/HPI.
4. Use of scoring tools with documentation (e.g., CHADS2, Depression scales).
5. Monitor for any drug toxicity or futility.
6. Preventive health (U.S. Preventive Services Task Force)—age, gender specific..
7. What to expect next.
8. Questions and education on critical topics.

Image: A multicolored bar captioned "Triage" shows the time spent on the following tasks:

• Follow up - 1 minute.
• Change - 1 minute.
• History/exam/Dx/Rx Plan - 3 minutes.
• Scoring - 2 minutes.
• Tox - 1 minute.
• Preventive Health -2 minutes.
• Education - 1 minute.

Average length of encounter (adult medicine: 11 minutes).

Average length of encounter (pediatric medicine: 14 minutes).

Average time spent on staying current/week: 35 min). [Image: Icon of an open book.]

Slide 9

To be clear: factor that impact an FDA decision and expert opinion may not necessarily equal a clinician's decision to prescribe

Adverse Event (AE) Signal:

• Bench/ex vivo.
• Animal.
• Frequency/severity/character of AE.
• Duration of effect.
• Magnitude of effect.
• Presence of prolonged prodrome.
• Confidence of attribution.

Marketplace Parameters:

• Need in the marketplace.
• Individual benefit.
• Route of delivery.
• History of safe utilization.
• Limits of tolerance (society vs individual).
• Projected utilization.

Slide 10

Are there Developer/Assessor Variables that Can be Evaluated by Users?—Obesity

• The country is becoming more obese.
• Almost every doctor takes care of obese patients.
• Congress and FDA believe that the lack of drugs is an unmet medical need.
  ("concerns over lack of availability of pharmacotherapies approved by the U.S. Food and Drug Administration (FDA) for treating obesity were expressed in September 2011 by the U.S. Congressional Committee on Appropriations. The committee stated "the lack of obesity medications is a significant unmet medical need." This committee directed FDA to develop a pathway by March 30, 2012, to support development of antiobesity treatments.")
• Experts generally indicated that both patient and clinician acceptance would be high for combination drug because the potential to eliminate long-term sequelae of obesity-related diseases is critically important.

Source: AHRQ Healthcare Horizon Scanning System Priority Area 10—Obesity.

Slide 11

Method

Case vignette: Ms. Lawrence is an obese businesswoman that is interested in making a change; getting tired, little time and has heard there may be some "new options."

Image: A faceless figure of a woman is shown. An arrow points from this figure to three text boxes, each labeled "More Information." Another arrows points from "More Information" to a group of ambiguous multicolored figures standing together; each figure has an empty "speech balloon" over its head. A third arrow points from this group to the text "clinical management opinion."

Slide 12

Deciphering the Crystal Ball in Obesity: Can you predict the future of care?

• 6 Question Survey.
• Posted 8/25/2012.
• Results Through 9/7/2012.
• 230 Respondents.
• 98% Respondents Physicians.
• 84% >10 Obese Patients/Month.

Images: A graph shows the cumulative number of respondents to the survey increasing from 1% on August 25, 2012, to 100% on September 7, 2012. A pie chart captioned "What is your profession?" shows that 98% of the respondents were physicians and 2% were categorized as "All other." A second pie chart captioned "How many obese patients do you encounter on average each month?" shows that 84% of the respondents encountered more than 10 obese patients per month; 8% encountered 6-10 obese patients, 5% encountered 3-5 obese patients, 1% encountered 1-2 obese patients, and 2% encountered none.

Slide 13

Who Provides Care to Ms. Lawrence?

Images: Two pie charts are shown. The first shows national encounter estimates for 2009 by physician specialty for women with a BMI of 40 over between the ages of 45 and 55:

• Primary Care: 53%.
• Surgery: 14%.
• Ob/Gyn: 9%.
• Cardio: 2%.
• All Other: 22%.

(n = 5,553,813.)

The second pie chart asks "Can you predict the future of care?" The following responses are shown by physician specialty:

• Primary Care: 50%.
• Surgery: 25%.
• Ob/Gyn: 1%.
• Cardio: 3%.
• All Other/Unknown: 21%.

(n = 230.)

Source: United States Department of Health and Human Services. Centers for Disease Control and Prevention. National Center for Health Statistics. National Ambulatory Medical Care Survey, 2009 [Computer file]. ICPSR31482-v3. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2011-11-17. doi:10.3886/ICPSR31482.v3.

Slide 14

Who is Ms. Lawrence?

National Encounter Estimates, 2009 Women with a BMI ≥40 Between 45-55 y.o.

Top 10 Drugs

Drug	Estimated Encounters
Metformin	680,071
Levothyroxine	387,678
Lasix	363,407
Lisinopril	360,204
Vytorin	358,967
Cymbalta	324,692
Nexium	318,056
Lipitor	317,137
Percocet-5	295,044
Benicar	288,628
All Other	16,112,833

Top 10 Diagnoses

Diagnoses	Estimated Encounters
Hypertension	1,056,193
Diabetes Type II	949,671
Obesity	325,396
Backache	240,515
Routine Medical Exam	224,340
Routine Gyn Exam	201,090
Postsurgical States	198,707
Lumbago	191,458
Myalgia and Myositis	179,426
Disc Degeneration	177,972

Images: Two pie charts are shown. The first shows national encounter estimates for 2009 for women with a BMI of 40 over between the ages of 45 and 55 by race:

1. White: 65%.
2. Black: 14%.
3. Asian: 0.6%.
4. Unknown: 19%.
5. All Other: 1%.

The second pie chart shows encounter estimates for 2009 for women with a BMI of 40 over between the ages of 45 and 55 by region:

• South: 41%.
• West: 18%.
• Midwest: 28%.
• Northeast: 13%.

Source: United States Department of Health and Human Services. Centers for Disease Control and Prevention. National Center for Health Statistics. National Ambulatory Medical Care Survey, 2009 [Computer file]. ICPSR31482-v3. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2011-11-17. doi:10.3886/ICPSR31482.v3.

Slide 15

Deciphering the Crystal Ball in Obesity
Assessing Time on Market

• Case Summary:
  • Ms. Lawrence is a 50 year old morbidly obese woman (BMI = 41) with a 5 year history of hypertension and a 3 year history of type 2 diabetes. She is a busy executive and travels frequently, does not have a diet or exercise plan and is easily fatigued. She is current on a regimen of metformin and insulin glargine (high dose) as well as losartan with good glycemic (HbA1c = 7.2) and blood pressure control (averaging 120s/80s mm Hg). Because of her very busy lifestyle, increasing fatigue and poor self image, she seeks your recommendations on weight loss options.
• Labs:
  • Hg: 12.1 g/dL, LDL: 125 mg/dL, HbA1c: 7.2, LFTs: AST—22 IU/L ; ALT—27 IU/L, Thyroid studies normal.

Image: A questionnaire is shown concerning the treatment of the sample patient, Mrs. Lawrence. The physician is asked to recommend on a scale from 1 to 5 (least recommended to highly recommended) the following treatment options: a recently approved pharmacotherapy with no known adverse events, a clinic trial for a new surgical procedure that may have less adverse outcomes than the current gastric band procedure, a recently approved combination antiobesity agent of existing medications which have some known adverse events, and gastric band surgery.

Slide 16

Time on Market: Recommended Therapy
Percent of Respondents Answering 4 or 5 (Recommend)

Specialty	New Combo Therapy	New Therapy	Established Surgical Procedure	New Surgical Procedure
PCP	31.3%	24.1%	39.5%	21.7%
Surgeons	13.5%	18.4%	28.9%	43.6%
Endo	53.8%	50.0%	16.7%	16.7%

PCP: surgery > RxComb > RxNME
Surgeons: new surgery > established > Rx
Endo: RxCombo > RxNME > surgery

Image: A line graph shows the percentage of respondents recommending each treatment option. 83 Primary care physicians responded; 39 surgeons responded, and 12 endocrinologists responded. The data is shown in the table above.

Slide 17

Deciphering the Crystal Ball in Obesity
Impact of Nonclinical Data

• Case Summary:
  • Mrs. Lawrence is very engaged during the clinical encounter, and after hearing your recommendation she decides that a surgical procedure is not a treatment option she wants to consider now. She wants more information about the two new drug treatments you have brought to her attention. After reviewing more of the product label you learn that one of the drugs had a significant animal safety signal which the other did not.

Image: A questionnaire is shown concerning the treatment of the sample patient, Mrs. Lawrence, after she has decided against surgical options and in light of new animal signal information about one of the medications. The physician is asked to rank a scale from 1 to 5 (least recommended to highly recommended) the following options: waiting 5 years until more safety data is acquired is prudent; the animal signal is concerning and too challenging to be disregarded; the animal signal is not concerning.

Slide 18

Non-clinical: Recommended Therapy
Percent of Respondents Answering 4 or 5 (Recommend)

Specialty	Animal Signal Concerning	Wait 5 Years	Animal Signal Not Concerning
PCP	31.1%	33.8%	21.6%
Surgeons	33.3%	34.3%	13.5%
Endo	25.0%	25.0%	72.7%

There is agreement between PCP/Surgeon on the need for caution re: animal signal; not so for endocrinologists.

Image: A line graph shows the percentage of respondents ranking each. 74 Primary care physicians responded; 37 surgeons responded, and 12 endocrinologists responded. The data is shown in the table above.

Slide 19

Deciphering the Crystal Ball in Obesity
Assessing Clinical Trial Subject Cohort 

• Case Summary:
  • Ms. Lawrence doesn't want to wait 5 years and is eager to do something as soon as possible. You discuss the data available in both product labels and show her that the non-combination product had more data on women in her BMI category, as well as in her age range, than the other product label did (albeit with slightly lower efficacy).

Image: A questionnaire is shown concerning the treatment of the sample patient, Mrs. Lawrence, after she has decided against waiting 5 years and the data available on both product labels has been discussed. The physician is asked to rank a scale from 1 to 5 (least recommended to highly recommended) the following options: recommend the medication from clinical trial subject cohort that studied patients similar to Mrs. Lawrence; recommend lifestyle change; clinical trial subject cohort has no bearing on recommendation.

Slide 20

Clinical Trial Subject Cohort: Recommended Therapy
Percent of Respondents Answering 4 or 5

Specialty	Trial Subj Cohort Important	About Lifestyle Change	Trial Subj Cohort Not Important
PCP	44.4%	59.7%	24.2%
Surgeons	26.9%	48.1%	26.9%
Endo	60.0%	60.0%	40.0%

Endo > PCP > Surgeon consider trial subject cohort important

Image: A line graph shows the percentage of respondents ranking each. 63 Primary care physicians responded; 27 surgeons responded, and 10 endocrinologists responded. The data is shown in the table above.

Slide 21

Conclusions

• Different "crowds" have different responses to the same data.
• This disparity in opinion regarding anti-obesity interventions is present and can be characterized—sources are not here assessed (e.g., variable information burden by specialty).
• "Horizon scanning," using a group of experts, could be augmented with larger technology user groups.
• Similarly, "safety scanning" might be a useful, proactive means of assessing a technology’s market risk if the user community does not understand how to best use such a technology.

Slide 22

Medscape Team: Acknowledgements

This is new—using an education platform for a different purpose. Can this...

• Better assess tolerability of harm?
• Inform benefit/risk communications?
• Improve REMS development?

Collaboration welcome!

Thanks to the Medscape Team

• Alan Baldwin.
• Karen Overstreet.
• Cyndi Grimes.
• Linda Giering.
• Lisa Miele.
• Victoria Anderson.

Slide 23

Thank You

Carl N. Kraus, MD
ckraus@medscape.net