Intervention Research

Question 1

Purpose of intervention research

Answer 1

establishes efficacy and effectiveness, as well as safety, of an intervention in a specific clinical population

Question 2

Gold standard for intervention research

Answer 2

Randomized controlled trial

Question 3

Patient-oriented evidence

Answer 3

outcomes that the patient cares about; mortality, symptom improvement, cost reduction, and quality of life.

Question 4

disease-oriented evidence

Answer 4

intermediate, physiologic, or surrogate end points that may or may not reflect improvements in patient outcomes (BP, blood chemistry, physiologic function, pathologic findings)

Question 5

Intervention research study quality: level one (SORT)

Answer 5

SR/meta-analysis or RCT’s with consistent findings High quality individual RCT’s All-or-none studys

Question 6

Intervention research study quality: level two (SORT)

Answer 6

SR/meta-analysis of lower quality clinical trials or of studies with inconsistent findings Lower quality clinical trial cohort study

Question 7

Intervention research study quality: level three (SORT)

Answer 7

“expert opinion”

Question 8

Randomized assignment

Answer 8

-Subjects have equal probability of being assigned to intervention “arms” -procedure: random number table coin toss computer generated **Concealed allocation!!** -This removes selection bias however it does NOT guarantee equivalent groups. -Do not confuse this with random sampling; this just gives everyone an equal chance of being assigned to a treatment group. -should take care of confounding variables

Question 9

allocation concealment

Answer 9

-Clinician that is performing the intervention is not making the decision about who is getting the intervention. -Occurs before randomization. -treatment assignment is done by research administrators not the clinicians. -Clinician DOESN’T have an option on what intervention they are performing on their patient. **sealed envelope**

Question 10

Blinding

Answer 10

-Almost always possible to blind the outcome assessment–> if this is done then it is a strength of the study.

Question 11

Intention-to-treat analysis

Answer 11

-Based on initial treatment intent -Participants are analyzed in the groups to which they are assigned at the start of the trial -“Crossing over” to other treatment(s), if it occurs, is ignored -Follow-up data from dropouts must be imputed -Imputing their data: “if they hadn’t dropped out they would’ve had this number”—last observation carried forward throughout the study rather then leaving the spot blank. (this is the easiest but not the best way to do it) -Contrasts with “per-protocol” analysis= analysis based on treatment actually received.

Question 12

protocol analysis

Answer 12

-when people cross out or drop out, their info isn’t imputed. -analysis is based on the treatment received.

Question 13

adequate follow up

Answer 13

-atleast 80%; less than 80 is a weakness more than 80 is a strength. -keep track of how many people dropped out -how many were eligible-> how many were randomized-> how many remained throughout the entire trial -want equal follow up across the treatment groups (ex: dont want 30% follow up in one and 90% follow up in the other)

Question 14

Key element of intervention research: were the group similar at the start of the study?

Answer 14

-They should be -should be a table that tells you thr demographics of each group -if the groups arent even then this should be accounted for in the statistical analysis

Question 15

Power

Answer 15

-Larger the sample size–> greater power -power is an issure before the study, or after if no treatment effect was found -study should analyze how many subject is needed to have adequate power before the start of the study–IRB generally requires this -power should be more than 80% to be considered significant

Question 16

Criteria for a high qulity RCT

Answer 16

-allocation concealment -blinding -groupd are equivalent at the start of the trial -intention-to-treat analysis -follow up >80% -adequate statistical power (sample size) ** if they have it, its a strength of the study **if they don’t, its a weaknes of the study

Question 17

Critical Appraisal: Purpose and Methods

Answer 17

Purpose: ID the purpose of the study Methods: what was the design? were valid and reliable outcome measures used for the primary outcome?–>how was this discussed in the article??

Question 18

Critical appraisal: Validity of the methods (SORT criteria)

Answer 18

-randomized assignment? allocation concealment? -masking? -intention-to-treat analysis? -at least 80% follow up? -similar groups at the start of the trial? -adequate power (>80%)?

Question 19

Critical appraisal: results

Answer 19

-what were the findings pertaining to the primary outcome? how were these determined? -were they clinically important? effect size, number needed to treat, minimal detectable change, minial clincally important difference

Question 20

Critical appraisal: discussion and conclusions

Answer 20

-are the conclusions justified? -limitations? -do these results apply to my patients? -do the results effect how i treat my patients? -summarize strengths and weaknesses using SORT

Question 21

Indicies of clinical significance

Answer 21

-effect size -number needed to treat (NNT) -minial detecable change (MDC) -minimal clinically important difference (MCID)

Question 22

Effect size: within group

Answer 22

-calculated using: Pre-test group mean score; post-test group mean score; standard deviation of pre-test scores

(pre-test mean - post-test mean)/ (pre-test standard deviation)

-assumes a statistically significant effect

Question 23

Effect size: between group

Answer 23

-calculated using: treatment group mean change score; control group mean change score; standard deviation (of control group vs. pooled)

(treament mean change - control mean change) / (standard deviation (control))

-assumes a statistically significant effect

Question 24

Interpreting effect size

Answer 24

“d”=effect size

trivial= <.20

small=.20-.50

moderate= >.50-.80

large= >.80

Question 25

Number needed to treat (NNT)

Answer 25

• number of pts who need to be treated for one additional (dichotomous) outcome
• how to calculate NNT: Determine “risk” (or event rate) of the outcome in the experimental and control group.

RISK (ER)= (# with outcome)/(total # in the group)

Question 26

minimal detectable change

Answer 26

• how much improvement in the (numeric) outcome falls within the “margin of error”
• MDC is based on the standard error of the measure (SEM) = variation due to change
• must detect change beyond 2 SEMs to conclude that change is not due to chance

Example: gait speed after hip fracture

SEM= .04 m/s

MDC=(1.96)(SEM)=(1.96)(.04)=0.08 m/s

Question 27

minimal clinically important change

Answer 27

• how much of the improvement in the (numeric) outcome is meaningful to the pt?
• based on one or more sources:

SEM

patient self report

expert opinion

ROC curve analysis

Example: gait speed after hip fracture (n=92)

MCID (by expert consensus and ROC curve)= 0.10 m/s

Question 28

RCT: pretest-posttest control group design

Answer 28

Question 29

RCT multigroup design

Answer 29

Question 30

RCT longitudinal design (repeated measures)

Answer 30

after randomizaion there is stricly a group that gets treatment and a control group. time and treatment are “factors”. Treatment effect is determined by the interaction between time and treatment. outcome is measure at baseline, immediately post intervention, 4 weeks, 6 months, and one year post intervention.

Question 31

RCT Cross over trial

Answer 31

Groupd are randomized into a treatment or control group. they get their treament and then following a “washout” period they switch and recieve the alternate treatment. Outcome is measured before and after each threatment session

Question 32

observational (cohort) design

Answer 32

-pragmatic clinical trials: comparative effectiveness research; practice-based research; clinical practice improvement research

Question 33

Cohort studies for interventions

Answer 33

subjects classified by treatment status

• prospective (concurrent): subjects followed into the future for outcome (disease) development
• Retrospective (non-concurrent): typically uses previously recorded info i.e. medical records

Question 34

Cohort studies for interventions: controlling confouding variables

• absent randomization, “control” is achieve statistically from stratification and/or multivariate analysis
• variables typically controlled for include: age/sex, co-morbidities, medical history and prior level of functioning, baseline levels of the outcome measure. other variables relevant to the question

Question 35

Control of confounding

Answer 35

distortion of the treatment-outcome relationship due to other variables (confounders)

• in RCT’s confounding is largely controlled by randomization
• in observational designs, confounding is controlled statistically