Subject: Family and Community Medicine Topic: Research Design 3 Lecturer: Dr.

Sosa Date of Lecture: August 3, 2011 Transcriptionist: Mating Cats =^.^= Pages: 5

RANDOMIZED CONTROLLED TRIAL (RCT) ´ A controlled experiment to assess the safety and efficacy of treatments for human diseases and health problems in which, treatments are assigned at random.

Control for confounders Prevent selection bias

Hierarchy of evidence for cause-effect relationships

GENERAL FEATURES: ´ Direct control of the investigator in the assignment of subjects to the comparison groups (manipulation of exposure) -Treatment - Control 1. Procedure and analysis follow the cohort study -prospective -what will happen after certain time? - analysis -comparison of incidence of outcome in the treatment and control groups 2. Random allocation to the comparison groups is the hallmark of an experiment The best study that will give best evidence will be RCT Diseases that can t use RCT: -fatal diseases -rare diseases TYPES OF EXPERIMENT 1. According to unit of randomization o Clinical trial often used with drugs/ new treatments o Community trial 2. According to purpose o Preventive/prophylactic trial *Reduction of risk of disease i.e. reduce recurrence of stroke o Therapeutic trial *Reduction of symptoms *Prevention of recurrence *Reduction of risk of death I.e. new drugs for prevention of skin cancer; new drugs for treatment of MI

- Quasi experiment: no randomization/ no control/ all will have intervention/ all with medication - True experiment o o with RCT EXPERIMENTAL STUDY

3. Strongest evidence for testing hypotheses of causation Why are RCT s important? ´ Results of RCTs provide the most secure basis for valid causal inferences

1

SY 2011-2012

3. According to phase Phase 1 Purpose Safety, dose range, side effects, form Effects, Side effects Effects, side effects, comparison with standard drugs (this is when you start your RCT) Adverse effects with long-term use (most patients needed) Subjects Healthy

Issues in the Selection of the Study Population o Inclusion and exclusion criteria should be defined Inclusion criteria: specific signs/symptoms, x-ray, PCR (+) Exclusion criteria: very sick, with other complications, pregnant Eligible sample population are those who passed the inclusion criteria Components of a Clinical Drug Trial: Experimental study: Sick o A research method in which one group of subjects receives a new treatment (e.g., drug, device, surgery), and they are compared with one or more other group(s) who receive different standard treatments or placebo. These groups are studied over the same time period using the same measures of safety and effectiveness. Experimental group: The group of patients that receives the drug under investigation. Control group: The group of patients that receives a different type of treatment, either a traditional one (already approved and used in therapy) or no treatment (a placebo).

2 3

Sick Sick

4

Schematic Diagram: o

o

o

N sample- all adult patients with TB S sample population must fulfill inclusion criteria before included in the sample population, this will be your elligible population R randomize and divide them to quadruple/triple; then after some time observe how many will have microbiological cure, how many will not how many @ triple, how many @ quadruble etc. Difference with cohort and others: Randomization The Analytic Experimental Research Designs Steps: o Select the study population o Allocate randomly to treatment or control groups o Determine/assess outcome status o Compute for measure of association o Interpret the measure of association ( The last 2 are methods of analysis)

Placebo: o An inactive form of treatment, usually an inert sugar pill, received by patients in the control group. The use of placebo in most clinical trials is declining due to ethical concerns.  (I.e. if standard treatment is present, it is not wise to use placebo, like antiepileptic agents which already have std treatment. The effect of the new drug must be compared with std (Phenobarbital)

o 

Cant give in rare/life threatening diseases i.e. breast cancer open

2

trials must be used where the px know what they are taking o I.e. Immunoglobulin for dengue-- Not only placebo is given but with other treatment (plasma effusions, blood transfusions, control of shock,  In this particular example standard treatment was given to both groups however, the immunoglobin is given only to one group . This is okay because the std treatment is given to both

***This is where bias of subjects develops. Ie the px know that the pill is only sugar result to negative effect on subject even at treatment group; if the subject knows about the treatment- even at the placebo group, he/she experiences the effect Procedures which can t be blinded: -Surgical procedures -Orthopaedic procedures (Blinding subjects for these procedures are considered unethical because the risk is high!!) Blinding: o The process of ensuring that almost everyone involved in the drug trial is unaware of who is receiving the experimental drug and who is receiving a traditional drug treatment, or a placebo, throughout the duration of the study. But there are times when you need to unblind like when severe adverse effects or fatalities are happening If at placebo group you have to treat If at treatment group you have to unblind and give treatment Then, you may then include these subjects in withdrawal group (failure of treatment)

Issues in the allocation to treatment or control group y y Random Compliance to treatment group

treatment

of

the

y Randomization: o The process of assigning individual patients to different treatment groups in such a way that each patient has the same chance, equal to and independent of every other patient, of being selected for any particular study or treatment group. The idea is to make all study groups as equal as possible at the beginning of the experiment this allows you to give the subjects equal chances to be in a part group; Investigators shouldn t know between treatment A and B to prevent selection bias o

o

y Use the same methods of ascertainment for treatment and control groups

Single blinded- subject doesn t know, observer and analyst know Double blinded- subject and observer doesn t know; only the analyst knows Triple blinded- subject, observer, analyst don t know 3

Possible reasons for noncompliance in RCT s o o o o o o o Misunderstanding of instructions. Inconvenience of participation. Side effects of treatment. Cost of participation. Forgetfulness. Disappointment with results. Preference for another treatment.

Issues in the assessment of outcome status o o Use the same method of ascertainment for treatment and control groups Minimize the follow-up bias (The intention to treat and evaluation must be equal for both groups. eg. The control group must be treated the same way as the exposed group)

Compliance with treatment: o The failure of patients to take drugs as prescribed is called noncompliance. Investigators cannot be certain that a treatment is or is not effective unless they can be assured that the treatment group is actually receiving the treatment. How will you if your subject is compliant? --The patient can tell you directly (Hindi ko po iniinom, sumasakit po kasi tiyan ko eh ) o How to ensure compliance? -- Give them allowance; this is ethical as long it is not too much and just enough to comply! (Minimum is P500.00/ patient) -- Ask for the drug blister to make sure that he/she drank it -- Ask for the vial before you give another set -- The subject has the right to withdraw anytime during the treatment, just state in the report the reason why s/he cannot be included in the AE of the drug. Do not forget to treat the patient if his/her reason for withdrawal is the occurrence of side effects.

o

o

Ethical issues in RCT s o Safety of the proposed treatment or intervention o Use of placebo in the control group o It is unethical to not treat and keep giving placebo on patients having side effects/ disease during the study. o Give standard treatment if it is available. o The safety of the patients is most important. o Informed consent o Patient s supply on the whole duration of the study o eg. It is unethical to give the patient 1 month worth of supply when the duration of the study is 2 months. o The patient should get all necessary support. Ethical considerations occur in both directions: o Ethics of introducing new procedures to individuals where these procedures might harm them. o Ethics in withdrawal of interventions from individuals when these interventions might be of benefit to them. SOME ETHICAL QUESTIONS o Is it proper to withhold from any patient a treatment that might give him benefit? o Is the proposed treatment safe for the patient? o Is it ethical to use a placebo? o Is it proper for the trial to be in any way masked? TO SOLVE SOME ETHICAL ISSUES 4

o

o

o

o

The trial should have a protocol which is submitted to an independent committee for approval. Before a trial can be carried out, the consent of the subjects to participate must be obtained (informed written consent). It is important that each participant should receive an explanation of the trial; subjects must be informed that they may be assigned to either the treatment or control groups. The risks of having the treatment must be explained; the possible benefits of the treatment must also be explained. None of the treatment options included should be known to be inferior to another based on previous randomization studies, & if a standard treatment regimen exists, it should be used as the control.

Incidence of outcome in the control Relative risk (rr) RR= new risk/ original risk RR= Rt/Rc RR: usually in decimal <1.0: treatment is beneficial =1.0: treatment has no effect >1.0: treatment is harmful

Absolute Risk Reduction (ARR) ARR= risk change ARR: usually in percent >0%: Treatment is beneficial =0%:Treatment has no effect <0%: Treatment is harmful ARR= Rc-Rt Rc: risk in the control Rt: risk in the treatment Relative Risk Reduction (RRR) RRR= risk change/original risk RRR= Rc-Rt / Rc RRR: usually in percent >0%: Treatment is beneficial =0%: Treatment has no effect <0%: Treatment is harmful

Measure of Association: Relative Risk

Example: express your change in weight, if it went down from 80kg ( Wc) to 60kg(Wt) ´ Absolute weight reduction: weight change= 20kg I lost 20kg ´ RR: new weight/original weight= 60/80=0.75 My risk is now 0.75 (I now weigh 75% of what I used to weigh) ´ Relative weight reduction: 20/80= 25% I lost 25% of my risk (I lost 25% of my weight)

Measure of Association Outcome Total + + A C -

Explanations for observations about association: Intervention B D A+B C+D A+B+ Total A+C B+D C+D o Interpretation: RR=1 ( ) association RR 1 (+) association R>1 (+)causative association RR<1 (+)preventive association Relative Risk= Incidence of outcome in the tx o o Truth Error y Random Error y Systematic error o The divergence of an observation from the true value due to chance alone BIAS: Consistent measurements distortions in

Controlling for Bias and Confounders o o Bias can occur at any stage of the study design: Choice of the study population

5

o o o o

Assessment of variables SELECTION BIAS INFORMATION/ MISCLASSIFICATION BIAS CONFOUNDING BIAS

DETECTION BIAS o Were persons responsible for outcome assessments unaware of the assigned therapy?

SELECTION BIAS ‡ During the selection of the study population ‡ The study population is not representative of the target population ‡ Due to: Inappropriate sampling method Non-responders or drop-outs Comparison groups are different in ways other than the variable of interest

Controlling for Bias and Confounding at the Design Stage y Preventing Selection Bias o Well-defined choice of population of appropriate sampling

INFORMATION/MISCLASSIFICATION BIAS ‡ During the data collection: Assessment of exposure status or outcome status ‡ Subjects are misclassified as to exposure or disease status ‡ Subjects : Recall Bias, Follow-up Bias (If the patient can/ cannot recall the effects of the medication to him/herself) ‡ Observers : Interviewer Bias (Variability in type of questioning makes the answers variable) ‡ Instruments/Methods: Measurement Bias CONFOUNDING BIAS o Occurs when a third variable confuses the relationship between the dependent and independent variables

o Use scheme

o Use of hospitalized controls (in case-control studies) y Preventing Information/ Misclassification Bias o Clearly written protocols: definition of terms Use of multiple sources of data to verify disease/exposure status Blinding Training of observers Use valid instruments, procedures and questionnaires

o

o o o

Confounder: o May also produce dependent variable changes in the y Preventing Confounding Bias o o Randomization Matching

o o

Associated with the independent variable A variable independently associated with the intervention or exposure Random allocation enables

Allocation Concealment y Centralised o Coded, identical containers On-site computer system Sequentially numbered, opaque envelop sealed,

o

o o

controlling for known confounders o random distribution of unknown confounders in treatment groups o

Attrition bias o Bias due to differences between groups in losses of participants from the study Withdrawals, deviation dropouts and protocol

Advantages of RCT s o Most efficient for investigating causality, because we can ensure that the cause precedes the effect (causal relationship between risk factors & disease). Possible confounding factors do not confuse the results.

o

o

6

o

Blinding of participants can reduce distortion in assessment of ourcomes

Disadvantages of RCT s o Decreased ability to generalize findings to other groups in the population(external validity). Problem subjects as the very young, the elderly & pregnant women who may have an adverse reaction to treatment are usually excluded from the trial. A long period of time often is required to reach a conclusion. A large number of participants may be required. Financial costs are typically high. Ethical concerns may arise. Problem of attrition (dropouts).

o

o

o o o

Question 1 Is quadruple therapy better than triple therapy in microbiological cure of TB among adult TB patients of UMC in 2011? Components of Good Research Question P E/I C O M - population (TB patients) -exposure/ intervention (quadruple therapy) -comparison (where to compare: triple ) - outcome (microbiological cure) - methodology (what design? RCT) ----------End of transcription--------Come to me all who labor and are heavy laden, and I will give you rest. Matthew 11:28

7

Sign up to vote on this title
UsefulNot useful