L10 RCT Flashcards
LO1: def and desc purp of clinic trials. Comp to other epidemiol studies.
-clinic trial= planned exp in pts, designed to elucidate most approp meth tx for fut pt with a cond. not for benef pp. new tx vs stand.
-purp- prov reliab ev of tx efficacy and safety.
Efficacy= (int val) abil of h/c interven to impr health of defined grp under specif conds (effectiveness ext).
Safety= (ext) abil interven NOT to harm defined grp in specif conds.
For fair comp trials must be: repro in exp conds, controlled comp of intervenes, fair and unbiased w/o confounding. Are subj to rand var- diffs obs in small trials und 1000, more prone to chance.
Clinic trial= PIII comp stand tx. Und 100K pt us.
Interven= prophylactic, diag, or therap gas/devices/regimens/proced etc.
LO2: expl disadvantage of non RCTs and use of hx controls.
-non RCTs= int alloc pts receiving new tx to comp with grp receiving stand. BUT alloc boas- by pt, clinician, investigator. Form of selec bias- once in, Sep eq and fairly. Norm selec bias- into study, not repres. Confounding- kn and unkn.
-why use RCT- why not info from prev pt on stand tx, why not give all pt new and comp info with hx controls-
Prob with comp hx controls- for stand tx grp:
Selec often less well def, less rigorous, diff meas. Treated diff from new tx grp. Less info about poten bias/confound. Diffic to adj for confound.
So they us overestim benef of new tx.
LO3: outl steps inv in RCT. Def factors, conduct trials, comp outc.
-define- dis of int, tx comp, outc meas, poss bias and confound, selec criteria and eligib, pts excluded.
-conduct- Id pt source, recruit, consent, maint to comparab grps.
Random ain’t- alloc to tx fairly w/o bias or confound.
Blinding- foll up identic and aim maint all pp’s, min loss, max tx compliance. (ITT). Assess using same criteria.
-comp outc- is there diff. Could it be chance (stat sig). How big is obs diff (clinic imp). Is obs diff attribute to tx, or bias/confound.
LO4: desc suitab outc meas for clinic trials.
Reasons for pre def outc. Prim and sec outc. Types outc. Feats of ideal outc. Timing of meas.
-reasons for pre def outc- what, when, how meas. Prev data dredging and rep anal (Ie p vals). Protocol for data collec. Agreed criteria for meas and ass outc.
-prim outc- pref only 1, used in samp size calc.
Sec- other outcs of int, often incl s/e.
-types outc- pathophys eg tum size, horm lev, ECG change.
Clinic def eg mort, morbid, disability.
Pt foc eg QoL, psych wellbeing, soc wellbeing, satisfac. Less obj.
-feats of ideal outc- eg death best. Outc meas pre def.
approp and relev to pt, clinician, soc etc. Obj.
Valid and attributable- obs eff linked to tx.
Sensit and specif- meas meth detec changes acc.
Reliab and robust- outc measurab by diff peop in var settings sim res.
Simp and sust- meas eas carried out rep.
Cheap and timely- not too expens to meas, or long lag time.
-timing of meas- baseline meas relev factors- monit inadvert diffs btw grps. Monit outc dur trial at set pts- For poss eff eg 1 grp disadvantage, For AE eg indiv pt harm. Final outc meas- comp effs of tx’s. Outc should be meas by assessor blinded to tx status.
LO5: disc advs of rand alloc and blind to min confound and bias in estim tx effs.
-non rand alloc- all excep rand. Alloc pp’s to tx by a person, hx basis, geog locat, convenience, numeric order etc. =poten alloc (selec) bias and confound facs = unidentifiable diffs btw grps.
-rand alloc- min alloc bias- randomisation= each pp eq chance each tx. Min confound-randomisation= tx grps likely simil size and chars by chance eg age, sex. kn and unkn confounding.
-randomisation- coin toss, rand n tables/gen.
-what diffs in outc btw tx grps may occ-
Chance- randomisation= bal in long run ie large trial. Small trials may have unbal incomparable grps.
Pt knows their tx- diff behav, expectant new tx better etc.
Clinicians know tx- diff choice secondary tx.
Assessor know tx- diff appr and sensit to meas outc, use well def obj criteria.
One tx is act better.
-knowing tx alloc/open label- pt and investigator. May=bias as pt behav/expec. Clinic alt tx/care/interest (non tx eff). Investigator alt appr to meas and ass outc (meas bias) eg more precise.
-blinding- not tell which tx and somet outc.
Min alloc bias-ensures randomisation. Rand alloc w/o blinding still alloc bias.
Foll up in identic ways- sing blind=one of assessor/pt/clinician (us pt). Double= us pt and clinician/assessor (us same person). Triple.
Aim make tx app identic- app, taste, text, dosage, warnings. Use desig pharmacy for RCTs to label with code numbs.
Diffic to blind- surgical procedure, psychotherapy eggs antiD, alt med eg acupunc vs drug, lifest interven, preven prots, obv AE. Ask pp which think to check.
LO6: desc placebo eff, what placebo is, how it addresses placebo eff.
- placebo= inert subst looks, tastes, pack identic to drug.
- comp no tx- leaves 1 grp unsure of obs diff due to new tx of receipt of care.
- placebo eff- psych benef from care or being on new tx. Can=impr clinic one if new tx not effec or no tx given. Attit.
- aim of placebo- cancel out placebo eff that exists in the active tx.
- ethic implics- only used when no stand tx avail. Form of deceptive. All pp’s informed may receive placebo. US FDA req but comp to stand tx act more usef. Need justif use as opp to stand tx. Only use placebo when indiv would have only got that anyway- not stand tx.
LO7: disc how deal with losses to foll up and non compliance.
-losses- pt clinic cond may= rem-approp. May choose withdraw-unfort.
Min- make foll up practical and decr inconvenience. Give suffic clear info at start. Avoid coercion or inducements. Maint contact.
-non compliance- misund instruc, dislike, think not work, pref diff tx, CBA.
Max- simp instruc, check compliance, ask about effs and s/e, ask about compliance.
-iss for anal- never 100% foll up or compliance. Anal need acc for.
LO8: expl diffs btw explanatory and pragmatic trials and expl mean of inten to treat anal.
- is phys action of new tx better than stand OR is new new tx better in routine clinic prac.
- explanation trial- as treated anal- ignores those who don’t take it. Est phys potency only. 2 grps no longer rand alloc as non compilers prob sys diff to compilers in way rel to symp=confound and bias.
- pragmat trial- inten to treat anal- incl all. Anal acc to origin grp alloc. Keep randomisation=decr selec bias and confound. Comps likely effs of tx’ in routine clinic prac.
- as treated us give larger sizes of eff so researchers pref. ITT smaller more realistic eff sizes. Clinic trials should norm be anal by ITT.
LO9: disc ethic princips inv in medic research in hum subjects.
-Dec Helsinki 00- health pt 1st consid. Act in pt int when prov medic care may weaken phys and ment cond. trials req to kn tx effic and safet. At that stage unkn which best= clinic equipoise. Pt not necess receive best tx and poss harm. Randomisation= no one chooses. Loss auton part compensated for by benef of trial if Indic will work. Better outc whether drug or placebo as incr monit and care. But trials often burden of examin, investig, frew foll up, more info reqs.
-collec and Indic ethics- tension btw dr oblit to indiv pt and collec desire to prov best tx. RCTs test.
Collec- all pt should have tx that are tested for effic and safet.
Indiv- trial prim benef fur pt but justif volunteer if needs of indivs and pts are met by compliance with ethic princips gov indiv care:
Beneficence princip- do good. RCT not guarantee benef.
Non maleficence- do no harm. RCT may.
Autonomy- pt own decis mak proc. RCT rand alloc.
Justice- no discrimination. RCT place burdens and confer benef.
LO10: desc iss should be consid for clinic trial to be ethical. Ethic appr by committ.
-clinic equipoise- reasonab uncert/genuine ignorance about the better tx (incl non interven). Most new tx in trials as Indic benef.
Iss- is uncert/ignore by indiv clinician or sci comm.
what constits reasonab uncert.
How is ‘better’ def for indiv pt or soc as whole.
-sci robust- adresses a relev/clinic imp iss. Asks valid Q. Has approp study design and protocol address bias and confound. Has poten to reach sound conclus- min inabil to find clinic imp eff using samp size calc. Can justif use of Stand tx/placebo. Acceptable risk of harm vs benef. Prov is for maint safet. Arts for approp reporting/publication.
Justif as pursuit of kn or benef soc-better.
Are for indep monit to stop trial if adv/disadvantage 1 grp.
Data protec- fraud res/ public bias.
-ethic recruitm-
Inapp inclus of- pp from comm unlikely to benef, can’t afford tx after trial, pp high risk harm wrt benef eg preg, pp likely excluded from anal eg small sub grp.
Inapp exclus of- pp diff from ideal homog grp eg ethnic/sex/comorb, pt diffic to get valid consent from eg immig/ ment I’ll/ kid/prisoner.
-valid consent- knowledgable informant. Approp info verb and written, cooling off period, freedom to opt out. Informed pp. competent decis maker. Legit authoriser. Us all same person, somet guardian. Allow ask Q and give approp resp. Signed consent form ev but not=val consent if prev conds not met. No coercion/manip or perception of!
-voluntariness- req for val consent. No coercion Eg non access to best tx, lower qual care, disint by clinician. No Manip eg exploit emot state, distort info, finance inducement.
Pow of clinician over pt judgem= diffic.
Is iss of autonomy- if induce infl exert so poten pp acts unchar in non auton way=unethical.
LO11: desc role and func of research ethics committ.
-lot ethic iss, impact int and ext val. (Attribute effs to tx and gen to pop). Can decr abil to prod sci val res. Us incr val as often address iss that concern pts likely to benef from tx.
-NHS trust/PCT R and D office- decr risk to organisation IMV- research gov, finan managem, resource implics.
-research ethics committ- protec pt. dignity, rights, safet, wellbeing.
Focus partic on: sci design and conduct, rec pp’s, care and protec, confid, informed consent, comm consids.
Any study int pt, relatives, staff, biol mater, or info from NHS ref to NHS REC.
-moral obligation to participate as all benef.
Interp RCT
- clinic trial as repro, controlled, fair exp study.
- feats of suitab outc meas.
- randomisation decr alloc bias and confound kn and unkn.
- blinding decr alloc bias, non tx effs and meas bias.
- placebo elim placebo eff.
- min loss to foll up.
- max compliance.
- ITT to preserve rand and refl prac.
Conduc RCT
- collec vs indiv ethics in biomed research.
- iss in ethic.
- REC role.
RCT
- absol risk= more likely to benef than not. (maj cured).
Rel risk= better than stand tx (more cured than other).
RCT is an Experimental study= conds under direc contr of researcher. Us giv grp interven that would not have occ nat. Often tests effs of peop dis and inv comp to grp no tx. Comp gens ratio.
