BME 212 Quiz 13

randomization is essential for testing treatment efficacy

eliminates selection bias, balances groups with respect to many known and unknown confounding variables, and forms the basis for statistical tests

simple randomization

based on a single sequence of random assignments, for instance, flipping a coin. problematic with small sample sizes (there will be differences simply by chance), resulting in unequal number of subjects (and confounding variables may be different at baseline). works well for n>100. For n=40, only 37% probability of similar sample sizes

block randomization

blocks could present sets of subjects that become available at different time points. subjects within blocks will be randomized but such that the groups are balances. even though balance in sample size is achieved groups are rarely comparable for specific covariates (confounding variables) potentially introducing bias statistical analyses correcting for unbalanced covariates can be applied but are not ideal

stratified randomization

addresses the need to control and balance the influence of covariates. one to two specific covariates that can impact the main outcome variables will be specified. blocks for combination of covariates are generated and subjects are assigned accordingly. then simple randomization assigns blocks/subjects groups. very useful for small sample size studies but may become complicated when attempting to control many covariates

covariate adaptive randomization

similar to stratified randomization in that it aims to keep covariates balanced between groups. it does not use blocks but instead uses the method of minimization by assessing the imbalance of sample size among several covariates

matched pairs

popular because of its simplicity. subjects are divided into pairs by matching each participant with their closest pair regarding some confouding variable(s). within each pair the subjects are randomly assigned to treatment and control groups

intention to treat analysis

once subjects were randomized and start on a regimen, they will be included in the analysis

will have to be performed for all clinical studies

per protocol analysis (on treatment)

when they stick with a regimen (per protocol)

optional; exclusion criteria can be changes by investigators but need to be justified

al subjects will be analyzed

regardless of whether they complied with the treatment they were given

intention to treat preserves randomization

randomization ensures that both measurable and unmeasurable factors will balance out on average. this prevents many types of bias that can occur in a non-randomized trial

intention to treat analysis is especially important for

medications that are difficult to tolerate. if you exclude noncompliant patients, you are ignoring the influence of poor tolerability on the efficacy of a treatment

oer protocol analyses are typically applied only

after the intention to treat analysis has been completed (as additional analysis)

a per-protocol analysis that excludes non-compliant patients may

produce a study population that is healthier than the patients that you see

negative effects of treatment that lead to non-compliance

are ignored

groups defined by compliance may not represent

the practical impact of the treatment (which doctors may be interested in)

per protocol analysis determines the impact

of the treatment if subjects follow directions (are compliant)

per protocol analysis gives researchers a lot of flexibility

for instance, different degrees of non-compliance could be defined

ethics protocol needs to be approved by the

institutional animal care and use committee (IACUC)

ethics for animal studies

is the welfare of the animals given full consideration?

are greatest lengths to minimize pain and/or discomfort taken?

are the animals anesthetized for the procedure?

are the animals given medication for pain?

is the number of animals justified?

are the minimal number of animals used to address a hypothesis?

are there alternatives means of addressing the hypothesis? in vitro studies?

using tissues from animals euthanized in other studies

is the research justified?

will is benefit animals/humans?

will the research ultimately benefit our understanding of life?

is the treatment/diagnostic/device being studied for the good of human kind?

research ethics- falsifying data

cooking your data

overlooking some data points

realizing that ignoring an outlier or that shifting/changing data will bring significance to an experiment

creating some data points

reusing same figure in different context

misrepresenting your findings

forgetting to report specific aspects of your methods

hiding some bad findings from your experiements

how common is scientific misconduct

scientific misconduct occurs on a scale

gold standard misconduct

open data, open code

questionable misconduct

sloppy statistics, peer review abuse, incorrect research design

misconduct

fabrication, falsification, plagiarism

prevalence of serious scientific misconduct is on the order of 1%

ultimately, science is self-correcting

conflicts of interest

personal interests, professional interests, financial interests, consulting for companies, relationship with fellow scientist

disclosure is critical-cannot be avoided but needs to be managed

must be disclosed to university, must be disclosed in scientific papers and seminars (ex. relavant patents, consultancies, or industry sponsors)

financial COI includes

compensation/remuneration, equity interests, royalty payments, sponsored travel