Transcript Your argument needs to leave the reader with the following

The Anatomy of a Protocol
•
•
•
•
•
Background
Research Question (hypotheses)
Design
Study Population
Measurement
– Predictors (intervention)
– Outcomes
– Confounders (Randomization integrity)
• <Procedure>
• Analysis
• Relevance/Contribution
Research Protocol
Page Allocation
Background and Model
4 1/2 pages
Research Question/Hypothesis
1/4 page
Design
1/2 page
Study Population
1/2 page
Measurement
4 pages
Predictors (intervention)
Outcomes
Confounders (randomization integrity)
Procedure
Analysis
1/2-1 page
Ethics/Relevance/Contribution
1/2 page
Your argument needs to leave the reader with the
following impression:
 This is an important question to answer (it may also
be interesting?)

We need to know the answer to this question
because…….
 it will have an impact on ____________
 it will change _____________________
 We do not know the answer to this question.
 The question can be answered by this study.
THE ARGUMENT PARADIGM





The Big Picture
how big is the problem?
burden of morbidity/mortality
impact on quality of life?
productivity?
cost of problem
__________________________________________________
Where does your question fit in?
Is it a logical next step?
 burden of illness
 determinants
 interventions
 cost
____________________________________
What will your
question answer that
isn’t known already?
Better mousetrap
Fill a hole
(no one knows and we need
to know because it could
make a difference)
Your question!
(no study to date has adequately
answered the question, the right
answer could change how we do
things, this study can/will solve this
problem)
The Research Cycle
Burden of disease
Causation
Determinants
Implementation
Efficacy
Efficiency
Effectiveness
After P Tugwell, 1985
PAGE ALLOCATION
The Big Picture
 How big is the problem?
 Burden of morbidity?
 Impact on quality of life?
 Productivity?
 Cost of problem?
__________________________________________________________
Where does your question fit in?
Is it a logical next step?
 Burden of illness
 Determinants
 Interventions
 Cost
_________________________________________________________________
What will your question answer that isn’t known already?
Better mousetrap
Fill a hole
(no one knows
and we need to
know because it
could make a
difference)
(no study to date has
adequately answered
the question, the
right answer could
change how we do
things, this study can/
will solve this problem)
GETTING STARTED
 Decide what you want to know from reviewing the literature
 sketch out the logic of your argument and find the related lit
OR
 read some review articles to get a handle on the area, the
assumptions, the unknowns
 Summarize as you go – the key elements?
 Having problems? Figure out why?
 question is not well defined?
 you have too many questions?
 your question is at the wrong part of the research cycle?
Appendix 1: A Summary of Studies Which Have Examined the Relationship between Benzodiazepines,Psychotropics and Injury in the
Elderly
Author
Year
Patients
Design (confounder adjustment)
Exposure
Risk Estimate
Risk of Hip Fracture in Relationship to Psychotropic and Other Drug Use
MacDonald {337}
1977
390 cases
case series
(no adjustment)
barbiturate
93% with hip fracture
used barbiturates as a
hypnotic
Rashiq {988}
1986
102 cases
204 controls
case-control
(age, sex match)
any drug
psychotropic
diuretic
0.42 (.26,.67)
0.84 (.41,1.73)
0.38 (.20,.72)
Ray {1120}
1987
1021 cases
5606 controls
nested
case-control
(design: sex, age, ace, location
antipsychotics
antidepressant
hypnotics-long
hypnotics-short
2.0 (1.6, 2.6)
1.9 (1.3, 2.8)
1.8 (1.3, 2.4)
1.1 (0.8, 1.6)
Taggart {334}
1988
282 cases
145 controls
case-control
(no adjustment)
sedatives
NSAID
1.08 (NS)
0.32 (p<.001)
Ray {984}
1989
4501 cases
24,041
controls
nested
case-control
(design: sex, age, index date)
Long-acting benzos
short-acting
benzos
1.7 (1.5, 2.0)
1.1 (0.9, 1.3)
Stevens {1123}
1989
173 cases
134 controls
case-control
(in analysis: age, sex, location,
widow, bodyweight, smoke,
dementia, stroke, arthritis,
diabetes, drugs)
benzodiazepine
tranquilizers
thiazides
nonthiazides
1.03 (0.6,1.8)
1.62 (0.6, 4.1)
1.11 (0.6, 1.9)
0.82 (.5, 1.4)
Grisso {1117}
1991
women ≥ 45 yr
hospital
admits
cases=174
controls=174
case-control
(design: age, hospital
analysis: age, LE dysfunction,
vision, stroke, body mass, alcohol,
meds)
Significant Risk Factors
LE dysfunction
vision
stroke
body mass
low, mid, high
1.9 (0.9, 3.8)
4.8 (1.4, 16.2)
4.5 (1.5, 13.5)
1.0
0.4 (0.2, 0.9)
0.2 (0.1, 0.5)
Functionalities, Use and Benefits from E-Rx systems
(Level 1=drug reference alone, Level 2=rx writer, no med history, Level 3=rx writer integrated with alerts, demographics, formulary info, allergies, Level 4=rx-writer and tracking of
meds, Level 5=rx-writer and connectivity between MDs, pharmacists and payers, Level 6=EMR integration)
Author/yr
Locale/journal
Level of System
Utilization Data
Ter
Wee/1993
UK (Br. J Gen
Practice)
Level 3
(rx-writer, demog from practice info system, enter disease for drug
list, drug interactions)
None
Proost/1992
Netherlands
(Compu Biol.
Med)
Level 1
(stand-alone drug reference that permits entry, storage and search for
patient drug hx, demographics, renal function, weight, and dose
recommendations for 180 drugs)
none
DeLeo/1993
U.S (Am J
Hosp Pharm)
Level 1
(self-administered, computerized patient medication history taking
structured interview-demographic, diseases, medication list,
compliance, symptoms, allergies, psychosocial-output format for
inclusion in chart)
ACCEPTABILITY
20 patients, mean age 41.7 yrs., mean completion time=40 minutes;
validated by pharmacist interview (no actual data), incomplete/wrong
data in 3/20
De
Zegher/1994
Belgium,
France, Italy
(Computer
Methods and
Programs)
Level 2
(standalone rx system that allows entry and storage of demographics,
and dx)
none
Purswani,
1995
U.S. (MD
Computing)
Level 1
(drug knowledge base-searchable monographs in hierarchical
structure)
ACCEPTABILITY
5/6 residents found easy to use without training
Puckett1995
U.S in-patient
(Am, J Hosp
Syst Pharm)
Level 5
CPOE system-with bar-coded verification at time of dispensing
(drug+ patient)-no DSS
PROCESS
Reduction in hospital wide annual rates of med errors (except wrong
patient) relative to doses dispensed by 0.17% to 0.7% (yr 1) and 0.5%
(year 2)
Berard,
1996
U.S in-patient
oncology
(Am, J Hosp
Syst Pharm)
Level 5-6
CPOE system for oncology drug management
PROCESS
Claims of no oncology drug errors but no method described
McCullin,
1997
U.S. in-patient
(Am, J Hosp
Syst Pharm)
Level 1-4
Computerized rules used to review doses dispensed for 35 drugs
relative to patient demographics and creatinine clearance, reports sent
to hosp pharmacists whp contacted MDs to fix problem
PROCESS
Of 28,528 orders, 2,859 (10%) had problems. Lower dose recommended
in 1,992 (70%), and higher dose in remainder, MDs contacted for 1163
(41%) of alerts. Altered dose in 868 (75%) cases
Gronroos,
1997
Finlandinpatient
Level 1
Drug interaction assessment system
PROCESS BASELINE
Among 2,457 inpatients, 326 serious interactions detected in 173 patient
(6.8%); calcium and fluroquinolones most common (n=66)
Example
•
•
Objective
To estimate risk of hospital
admissions for cardiovascular
and respiratory diseases
associated with PM10-2.5
exposure, controlling for
PM2.5.
Background