Title: Having an Impact in a Multi-Disciplinary Setting
1Having an Impact in a Multi-Disciplinary Setting
I am speaking as a private citizen and not as an
official FDA representative. The statements I am
about to make reflect my personal opinions and
are not necessarily supported or endorsed by FDA.
2Having an Impact
- For Statisticians, Methodologists,
Epidemiologists, - What Does Having an Impact Mean?
- Typical Challenges
- Practical Approaches
3Having an Impact Tale 1
- An investigator came to my office asking for
sample size requirements - I could never get a clear picture of the overall
situation - She put my questions aside, all she wanted was
the number - After I finally coughed up a number for her she
never got back to me - I admit I did not try very hard to follow up with
her either - My impression is that the number was too large
4Having an Impact Tale 2
- I was working as a biostatistician for a large
grant - The primary analyses specified in our grant were
not significant - In the exploratory analyses, the intervention
effects were significant - The investigator asked me to help draft a
manuscript based only on the exploratory results
- I explained the difference between exploratory
analyses and confirmatory analyses - After a long series of discussions, I finally
persuaded her and we agreed to report both the
primary analysis results and the exploratory
analysis results.
5What Does Having an Impact Mean?
- Promoting good statistical practices
- Promoting statistical thinking
- Contributing to the goals of the organization
- Contributing to the development of research
methodology - Other ideas?
6Tale 1 Lack of Involvement Limits Impact
- The statistician
- Encountered resistance to her questions (why?)
- Provided an answer that probably didnt satisfy
anybody
7Tale 2 Involved Statistician Has an Impact
- The statistician
- Supported the overall research objectives
- Explained his concerns and consequences in the
context of the study conclusions - Suggested an alternative approach
- Maintained a collaborative relationship with the
team
8What does it take for a statistician to have an
impact in a multi-disiplinary setting?
- Statistical Expertise
- Non-Statistical Skills
- Vision of Broad Role
Impact
Involvement
9Challenges to Having an Impact
Her solutions dont look very realistic.. We
dont understand the report or the relevance of
her input. Statisticians dont give us answers
that we can use. Well just make a decision
based on what weve always done in the past.
They didnt appreciate all of my hard work on
their problem. They didnt understand how my work
addresses their questions. Clients from this
field seem to think they can just ignore the
statistical evidence and make arbitrary
decisions.
Statistician
Client
10Challenges to Having an Impact
- Role of the statistician
- Communication skills
- Strategic issues
11Challenge 1 Role of the Statistician
- Helping clients think through their scientific
goals, working out details of study design and
implementation, and figuring out how to display
and interpret results are major tasks. In fact,
refining a good idea for a research study into
concrete specific aims and study design is one of
the most challenging parts of doing research, and
something with which an experienced statistician
can be a great help. Kevin Cain, The
Statistical Consultant, Fall, 2006
12Practical Approaches
- Graduate student education in statistics/methodolo
gy needs to include full involvement in actual
projects - Practicing statisticians Evaluate the level of
your involvement in projects
13Challenge 2 Communication Skills
Asking Good Questions
Talking About Statistics
14Practical Approach Improve Your Skills
1. So, what is your target population? 3.
Well, you mentioned this forty thousand. Now is
that undergraduate students on their own or is it
graduates and undergraduates? 5. OK, so are
you able to get a random probability sample of
this target population?
2. Uh, you mean the people we want to find out
about? 4. Well, its graduates and
undergraduates... 6.A what?
15Use Closed Probes for Specific Information
- How many nursing homes are in the database?
- Can you conduct interviews in person instead of
over the phone? - Did each surgeon evaluate the records of all 202
patients for risk of complications? - Can you obtain a matching set of records for
patients who did not use antidepressants?
16Use Open Probes for General Information
- Why do you need a seven-year reference period for
recalling product purchases? - How does a surgeon express the risk of
complications in the abdominal aneurism surgery? - How would clinicians view each of the outcomes
related to incontinence in the nursing home
patients? - What other factors might influence a patients
insulin dosage over the 14-year reference period?
17Use Concrete Paraphrase To Confirm Your
Understanding
- My understanding is that
- as a diabetic patient ages he tends to use a
larger dose of insulin. - we could develop a plan to conduct an in-person
interview with fewer subjects but with a greater
response rate. - clinicians view the use of a catheter as a
separate outcome not related to incontinence. - the computer program was developed independently
of the surgeons who evaluated the patients
records.
18Asking Good Questions
G
G
S
S
REQUEST FOR
CLOSED PROBE
E
P
E
P
INFORMATION
E
N
N
E
C
E
C
E
CONCRETE PARAPHRASE
I
I
R
R
OPEN QUESTION
F
F
A
A
I
I
L
L
C
C
CONCRETE PARAPHRASE
S
S
REQUEST FOR
G
G
P
INFORMATION
P
E
E
E
E
N
N
C
C
E
E
CONCRETE
I
I
CLOSED PROBE
R
R
PARAPHRASE
F
F
A
A
I
I
L
L
C
C
CONCRETE PARAPHRASE
OPEN QUESTION
19Talking About Statistics
Video examples of good and poor communication
skills
20Talking About Statistics
- Be prepared to communicate statistical
information in a variety of different ways - Be attuned to your client to pick up clues about
his/her learning preferences (visual, auditory,
kinesthetic) - Adapt your discussion accordingly, on the spot!
21Challenge 3 Strategic Issues
- Clients not aware of benefits of collaboration
- Lack of previous collaborative experiences
- Client discipline does not have a tradition of
involving statisticians - Institutional barriers
- Limited resources
- Position of statistician in the organization
22Practical Approaches
- Provide talks, short courses to clients
- Invite clients and statisticians to discuss
common methodological concerns - Identify clients who can serve as role models to
their colleagues for collaboration and innovation - Discuss with clients how to minimize
institutional barriers
23Having an Impact in a Multi-Disciplinary Setting
- In my opinion, statisticians can help facilitate
effective collaborative relationships with
clinical researchers by being proactive in being
included in research teams from the inceptions of
projects. In all situations, asking the
questions needed to gain a clear understanding of
the research and related clinical issues should
be considered a prerequisite to providing
assistance. - Collaborative relationships could be further
enhanced by training statisticians in
consultative approaches and methods along with
real-world practice providing consultation in a
variety of settings. Janet Powell, The
Statistical Consultant, Fall, 2006